{ "reviewDate": "2026-10-23", "methodology": "https://ethicalstack.ai/methodology/", "items": [ { "id": "claude-code", "name": "Claude Code", "category": "harnesses", "organization": "Anthropic", "website": "https://www.anthropic.com/claude-code", "summary": "Terminal coding agent that reads and edits a local project while routing requests to Anthropic services.", "scope": "The public repository contains the product wrapper/docs and plugins, but its license reserves rights; model and hosted service are separate.", "dimensions": { "energy": { "summary": "Claude Code does not publish service-specific energy or water data; the model and serving provider determine most inference impact.", "sourceUrls": [ "https://docs.anthropic.com/en/docs/claude-code/getting-started" ], "confidence": "low", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Claude Code requires a network connection for inference and supports Anthropic’s API plus Bedrock and Vertex deployments. That gives buyers a provider choice, but the docs provide no workload-level electricity, water, or emissions data. Choose a provider with regional energy reporting and compare the same model and task before treating one route as lower impact.", "argument": "Claude Code requires a network connection for inference and supports Anthropic’s API plus Bedrock and Vertex deployments. That gives buyers a provider choice, but the docs provide no workload-level electricity, water, or emissions data. Choose a provider with regional energy reporting and compare the same model and task before treating one route as lower impact.", "assessmentType": "argument", "score": null, "href": "/harnesses/claude-code/#energy" }, "training-rights": { "summary": "Claude Code is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://privacy.anthropic.com/en/articles/10023580-is-my-data-used-for-model-training", "https://privacy.anthropic.com/en/articles/7996868-is-my-data-used-for-model-training" ], "confidence": "low", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "Claude Code itself does not disclose a training corpus for its model; the model is Anthropic-hosted by default, with enterprise routes through Bedrock or Vertex. Anthropic’s consumer and commercial policies give different training rules, but neither page provides a source-level rights ledger. Check the exact account and provider terms, or use a separate local-model harness if you need to inspect the corpus before sending code.", "argument": "Claude Code itself does not disclose a training corpus for its model; the model is Anthropic-hosted by default, with enterprise routes through Bedrock or Vertex. Anthropic’s consumer and commercial policies give different training rules, but neither page provides a source-level rights ledger. Check the exact account and provider terms, or use a separate local-model harness if you need to inspect the corpus before sending code.", "assessmentType": "conditional", "score": null, "href": "/harnesses/claude-code/#training-rights" }, "privacy": { "summary": "Anthropic’s Claude Code data rules vary by consumer versus commercial account and opt-in/safety-review state.", "sourceUrls": [ "https://privacy.anthropic.com/en/articles/10023580-is-my-data-used-for-model-training", "https://github.com/anthropics/claude-code" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 6, "scoreReason": "Use Claude Code only after checking the account path: consumer, commercial, API, and ZDR terms differ, so a general “Claude privacy” claim does not fit every session. Turn off training where available and confirm where logs are retained for the exact provider route.", "argument": "Use Claude Code only after checking the account path: consumer, commercial, API, and ZDR terms differ, so a general “Claude privacy” claim does not fit every session. Turn off training where available and confirm where logs are retained for the exact provider route.", "assessmentType": "evidence", "score": 6, "href": "/harnesses/claude-code/#privacy" }, "openness": { "summary": "The product repository is public, but its license reserves rights; it is not open-source software.", "sourceUrls": [ "https://github.com/anthropics/claude-code", "https://github.com/anthropics/claude-code/blob/main/LICENSE.md" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 2, "scoreReason": "Claude Code’s public repository makes client code and plugins inspectable, but its license reserves rights and does not make the software open source. This access does not expose Claude models or Anthropic’s hosted inference service. Treat the editor and model as separate parts of the stack.", "argument": "Claude Code’s public repository makes client code and plugins inspectable, but its license reserves rights and does not make the software open source. This access does not expose Claude models or Anthropic’s hosted inference service. Treat the editor and model as separate parts of the stack.", "assessmentType": "evidence", "score": 2, "href": "/harnesses/claude-code/#openness" }, "labour": { "summary": "Company context: Anthropic. Anthropic’s 2025 Claude 4 system card and inbound-services agreement disclose use of paid data-labeling contractors and require crowd-work vendors to follow minimum worker-wellbeing expectations. Stanford FMTI found no public worker compensation, location, or top-source labor-practice outcomes, so the standards remain policy commitments rather than independently verified results. This is a record of the company behind Claude Code; it does not establish use of this particular product or release.", "sourceUrls": [ "https://www-cdn.anthropic.com/de8ba9b01c9ab7cbabf5c33b80b7bbc618857627/Model%5C_Card%5C_Claude%5C_3.pdf", "https://crfm.stanford.edu/fmti/December-2025/company-reports/Anthropic_FinalReport_FMTI2025.html", "https://www.anthropic.com/legal/inbound-services-agreement" ], "confidence": "medium", "unknowns": "Working conditions for the people building or supporting Claude Code are not established by this company-wide record. Current pay, vendor locations, worker exposure, and compliance/audit results are not published per major data source.", "suggestedScore": 4, "scoreReason": "Company-level assessment: Public vendor-wellbeing standards and contractor disclosure are meaningful, but Stanford FMTI found no compensation or worker-level outcomes for the principal data sources. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "conditional", "argument": "Public vendor-wellbeing standards and contractor disclosure are meaningful, but Stanford FMTI found no compensation or worker-level outcomes for the principal data sources. This is relevant when deciding whose business to support, while pay and working conditions for the team behind Claude Code remain unestablished. Ask for current worker outcomes for that team and its suppliers.", "assessmentScope": "Company record", "productArgument": "Stanford’s 2025 assessment says Anthropic uses paid contractors and data-work platforms, but finds its public disclosure incomplete on worker pay, locations, and protections. This concerns model-development labor rather than Claude Code’s software team. Ask Anthropic for current vendor-level worker standards and independent verification before relying on its model services.", "score": 4, "href": "/harnesses/claude-code/#labour" }, "military-contracts": { "summary": "Company context: Anthropic. Anthropic reports a two-year DoD CDAO prototype agreement with a $200 million ceiling, Claude in classified Palantir mission workflows, and Claude Gov deployments on AWS for national-security agencies. In March 2026, Anthropic said DoW designated it a supply-chain risk, narrowly applying to Claude use directly under DoW contracts, and that it would support a transition for warfighters. This is a record of the company behind Claude Code; it does not establish use of this particular product or release.", "sourceUrls": [ "https://www.anthropic.com/news/anthropic-and-the-department-of-defense-to-advance-responsible-ai-in-defense-operations", "https://www.anthropic.com/news/where-stand-department-war" ], "confidence": "high", "unknowns": "Deployment of Claude Code under these contracts is not established. These company statements do not establish the specific operational outcomes or independently assess the safeguards of deployments.", "suggestedScore": 3, "scoreReason": "Company-level assessment: The company documents classified defense use and a $200 million prototype agreement; deployments and guardrails are not independently assessed. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "The company documents classified defense use and a $200 million prototype agreement; deployments and guardrails are not independently assessed. This is relevant when deciding whose business to support. Deployment of Claude Code or this particular model release under those contracts remains unestablished; consider the company relationship separately from the product’s actual use.", "assessmentScope": "Company record", "productArgument": "Anthropic announced a two-year DoD prototype with a $200 million ceiling and public responsible-use limits. This is a direct company relationship, not proof that Claude Code itself was deployed. Buyers can weigh Anthropic’s provider relationship and ask which products and models were used.", "score": 3, "href": "/harnesses/claude-code/#military-contracts" }, "israel-palestine": { "summary": "We did not find Anthropic among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Anthropic is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Anthropic names AWS as its primary training and cloud provider and reports a large compute commitment and Amazon investment. This connects its business to a BDS pressure target without making Anthropic a separately named target; the announcement does not identify the host of every Claude request.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://www.anthropic.com/news/anthropic-amazon-compute" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": { "label": "Amazon partnership", "sourceIds": [ "anthropic-aws" ], "argument": "Anthropic names AWS as its primary training and cloud provider and reports a large compute commitment and Amazon investment. This connects its business to a BDS pressure target without making Anthropic a separately named target; the announcement does not identify the host of every Claude request." }, "businessContext": null, "score": null, "href": "/harnesses/claude-code/#israel-palestine" } }, "sources": [ { "url": "https://github.com/anthropics/claude-code", "title": "Official source repository or product page", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/anthropics/claude-code/blob/main/LICENSE.md", "title": "Official license or documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://privacy.anthropic.com/en/articles/10023580-is-my-data-used-for-model-training", "title": "Anthropic consumer model-training policy, including Claude Code", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/news/anthropic-and-the-department-of-defense-to-advance-responsible-ai-in-defense-operations", "title": "Anthropic Department of Defense prototype agreement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.anthropic.com/en/docs/claude-code/getting-started", "title": "Claude Code provider and network requirements", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://privacy.anthropic.com/en/articles/7996868-is-my-data-used-for-model-training", "title": "Anthropic commercial model-training policy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://crfm.stanford.edu/fmti/December-2025/company-reports/Anthropic_FinalReport_FMTI2025.html", "title": "Stanford CRFM 2025 Anthropic transparency assessment", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://www-cdn.anthropic.com/de8ba9b01c9ab7cbabf5c33b80b7bbc618857627/Model%5C_Card%5C_Claude%5C_3.pdf", "title": "Claude 3 Model Card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/legal/inbound-services-agreement", "title": "Anthropic Inbound Services Agreement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/news/where-stand-department-war", "title": "Where things stand with the Department of War", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/news/anthropic-amazon-compute", "title": "Anthropic: expanded Amazon compute agreement, April 2026", "kind": "primary", "checkedAt": "2026-10-23" } ], "href": "/harnesses/claude-code/", "overall": { "score": 3.8, "rated": 4, "total": 6, "coverage": 0.6666666666666666, "partial": true, "color": "red", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 15 } }, { "id": "codex-cli", "name": "OpenAI Codex CLI", "category": "harnesses", "organization": "OpenAI", "website": "https://openai.com/codex/cli/", "summary": "Open-source terminal coding agent that operates in a user-selected project and can use local or OpenAI-hosted execution.", "scope": "The CLI is a harness; model, account policy, and cloud execution are separate.", "dimensions": { "energy": { "summary": "OpenAI Codex CLI does not publish service-specific energy or water data; the model and serving provider determine most inference impact.", "sourceUrls": [ "https://github.com/openai/codex" ], "confidence": "low", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Codex runs its command loop in a terminal, but the default model call still goes to OpenAI. Local file access is not evidence that inference runs on the developer’s machine, and OpenAI publishes no Codex task energy figure here. If energy matters, use the lightest capable model and ask OpenAI for region- and workload-specific reporting.", "argument": "Codex runs its command loop in a terminal, but the default model call still goes to OpenAI. Local file access is not evidence that inference runs on the developer’s machine, and OpenAI publishes no Codex task energy figure here. If energy matters, use the lightest capable model and ask OpenAI for region- and workload-specific reporting.", "assessmentType": "argument", "score": null, "href": "/harnesses/codex-cli/#energy" }, "training-rights": { "summary": "OpenAI Codex CLI is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://help.openai.com/en/articles/7730893-data-controls-faq", "https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan" ], "confidence": "low", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "Codex CLI uses OpenAI models by default, and account settings determine whether prompts may be used to improve them. The product does not publish a full training-data rights ledger for those models. Choose an API or business route with the needed data controls, and request model-specific provenance if training rights matter to the decision.", "argument": "Codex CLI uses OpenAI models by default, and account settings determine whether prompts may be used to improve them. The product does not publish a full training-data rights ledger for those models. Choose an API or business route with the needed data controls, and request model-specific provenance if training rights matter to the decision.", "assessmentType": "conditional", "score": null, "href": "/harnesses/codex-cli/#training-rights" }, "privacy": { "summary": "Codex privacy depends on local versus cloud execution and the account’s data-control settings.", "sourceUrls": [ "https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan", "https://help.openai.com/en/articles/7730893-data-controls-faq", "https://github.com/openai/codex" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 5, "scoreReason": "The CLI can run commands in your environment, but model inference and cloud tasks still involve OpenAI services. For private work, choose local execution, confirm the applicable account’s training settings, and distinguish local files from cloud sessions.", "argument": "The CLI can run commands in your environment, but model inference and cloud tasks still involve OpenAI services. For private work, choose local execution, confirm the applicable account’s training settings, and distinguish local files from cloud sessions.", "assessmentType": "evidence", "score": 5, "href": "/harnesses/codex-cli/#privacy" }, "openness": { "summary": "Codex CLI repository is Apache-2.0; hosted models and cloud services remain separate.", "sourceUrls": [ "https://github.com/openai/codex", "https://github.com/openai/codex/blob/main/docs/license.md" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 9, "scoreReason": "The Codex CLI is available under Apache-2.0, so users can inspect, modify and redistribute the client. That license does not cover OpenAI’s hosted models or cloud services. The command-line tool and its inference backend have different access terms.", "argument": "The Codex CLI is available under Apache-2.0, so users can inspect, modify and redistribute the client. That license does not cover OpenAI’s hosted models or cloud services. The command-line tool and its inference backend have different access terms.", "assessmentType": "evidence", "score": 9, "href": "/harnesses/codex-cli/#openness" }, "labour": { "summary": "Company context: OpenAI. OpenAI’s supplier code sets broad labor standards but does not publish worker outcomes. TIME’s investigation reviewed OpenAI–Sama contracts, payslips, and interviews about 2021–22 Kenyan toxic-content labeling; workers described low take-home pay and distress, while Sama disputed workload/counseling claims and OpenAI said Sama managed pay and support. TIME later reported OpenAI said it was unaware of the conditions and no longer worked with Sama; this is a historical contractor case, not evidence of current API annotation conditions. This is a record of the company behind OpenAI Codex CLI; it does not establish use of this particular product or release.", "sourceUrls": [ "https://openai.com/policies/supplier-code/", "https://time.com/6247678/openai-chatgpt-kenya-workers/", "https://time.com/7012787/mophat-okinyi/" ], "confidence": "medium", "unknowns": "Working conditions for the people building or supporting OpenAI Codex CLI are not established by this company-wide record. The sources do not establish whether OpenAI uses other current annotation vendors, present worker outcomes, or the effectiveness of the supplier code across vendors.", "suggestedScore": 4, "scoreReason": "Company-level assessment: A documented past supplier episode includes worker reports of low pay and psychological distress, countered by Sama and OpenAI responses; the current supplier code has no public independent outcome audit. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "A documented past supplier episode includes worker reports of low pay and psychological distress, countered by Sama and OpenAI responses; the current supplier code has no public independent outcome audit. This is relevant when deciding whose business to support, while pay and working conditions for the team behind OpenAI Codex CLI remain unestablished. Ask for current worker outcomes for that team and its suppliers.", "assessmentScope": "Company record", "productArgument": "TIME reported that OpenAI contracted Sama in 2021 for safety labeling and described low pay and psychologically distressing material; OpenAI and Sama disputed parts of the account. That is relevant to OpenAI’s model pipeline, not proof about Codex CLI staff or current suppliers. Ask OpenAI for current vendor, pay, well-being, and grievance evidence.", "score": 4, "href": "/harnesses/codex-cli/#labour" }, "military-contracts": { "summary": "Company context: OpenAI. OpenAI publicly describes a 2026 classified Department of War agreement with cloud-only deployment, retained safety controls, cleared staff, and contractual limits. It separately says ChatGPT will reach GenAI.mil for unclassified work used by 3 million civilian and military personnel. Both are direct defense deployments; limits are company-reported and not independently audited. This is a record of the company behind OpenAI Codex CLI; it does not establish use of this particular product or release.", "sourceUrls": [ "https://openai.com/index/our-agreement-with-the-department-of-war/", "https://openai.com/index/bringing-chatgpt-to-genaimil/" ], "confidence": "high", "unknowns": "Deployment of OpenAI Codex CLI under these contracts is not established. Actual use, oversight effectiveness, and compliance with contract limits are not independently established.", "suggestedScore": 5, "scoreReason": "Company-level assessment: OpenAI reports cloud-only deployment, retained safety controls, cleared staff, and contractual limits; these are meaningful safeguards, but no independent check of enforcement is public. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "OpenAI reports cloud-only deployment, retained safety controls, cleared staff, and contractual limits; these are meaningful safeguards, but no independent check of enforcement is public. This is relevant when deciding whose business to support. Deployment of OpenAI Codex CLI or this particular model release under those contracts remains unestablished; consider the company relationship separately from the product’s actual use.", "assessmentScope": "Company record", "productArgument": "OpenAI announced a Department of War agreement for classified deployments and stated restrictions on autonomous weapons and domestic surveillance. This is a direct OpenAI relationship, not evidence that Codex CLI was the operational interface. Buyers can consider the parent-provider relationship and ask which models and tools served the contract.", "score": 5, "href": "/harnesses/codex-cli/#military-contracts" }, "israel-palestine": { "summary": "We did not find OpenAI among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "OpenAI is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. OpenAI’s April 2026 agreement names Microsoft as its primary cloud partner and a major shareholder, with revenue-share payments to Microsoft through 2030. That is a documented commercial link to a BDS target, not a separate official listing of OpenAI; local use of downloaded weights does not by itself establish a new payment to either company.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://openai.com/index/next-phase-of-microsoft-partnership/" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": { "label": "Microsoft partnership", "sourceIds": [ "openai-ms" ], "argument": "OpenAI’s April 2026 agreement names Microsoft as its primary cloud partner and a major shareholder, with revenue-share payments to Microsoft through 2030. That is a documented commercial link to a BDS target, not a separate official listing of OpenAI; local use of downloaded weights does not by itself establish a new payment to either company." }, "businessContext": null, "score": null, "href": "/harnesses/codex-cli/#israel-palestine" } }, "sources": [ { "url": "https://github.com/openai/codex", "title": "OpenAI Codex CLI source repository", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/openai/codex/blob/main/docs/license.md", "title": "Official license or documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan", "title": "Official product documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://help.openai.com/en/articles/7730893-data-controls-faq", "title": "OpenAI data controls and training preferences", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/our-agreement-with-the-department-of-war/", "title": "OpenAI Department of War agreement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://time.com/6247678/openai-chatgpt-kenya-workers/", "title": "TIME investigation into OpenAI-Sama data-labeling work", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/policies/supplier-code/", "title": "OpenAI Supplier Code of Conduct", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://time.com/7012787/mophat-okinyi/", "title": "Mophat Okinyi: TIME100 AI 2024", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/bringing-chatgpt-to-genaimil/", "title": "Bringing ChatGPT to GenAI.mil", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/next-phase-of-microsoft-partnership/", "title": "OpenAI: amended Microsoft partnership, April 27 2026", "kind": "primary", "checkedAt": "2026-10-23" } ], "href": "/harnesses/codex-cli/", "overall": { "score": 5.8, "rated": 4, "total": 6, "coverage": 0.6666666666666666, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 8 } }, { "id": "cursor", "name": "Cursor", "category": "harnesses", "organization": "Anysphere", "website": "https://cursor.com/", "summary": "Commercial AI coding editor and agent with local editing, cloud/background-agent modes, and selectable model providers.", "scope": "Privacy and retention depend on Privacy Mode, selected provider, and background-agent use.", "dimensions": { "energy": { "summary": "Cursor does not publish service-specific energy or water data; the model and serving provider determine most inference impact.", "sourceUrls": [ "https://docs.cursor.com/settings/api-keys" ], "confidence": "low", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Cursor lets users provide API keys for supported providers, while specialized features can continue to use Cursor’s built-in models. Those different routes make the energy boundary feature-specific, and the docs contain no task-level impact data. Check which provider serves each feature and compare its location and energy disclosure before choosing a plan.", "argument": "Cursor lets users provide API keys for supported providers, while specialized features can continue to use Cursor’s built-in models. Those different routes make the energy boundary feature-specific, and the docs contain no task-level impact data. Check which provider serves each feature and compare its location and energy disclosure before choosing a plan.", "assessmentType": "argument", "score": null, "href": "/harnesses/cursor/#energy" }, "training-rights": { "summary": "Cursor is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://docs.cursor.com/account/privacy", "https://docs.cursor.com/models" ], "confidence": "low", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "Cursor lets users bring provider API keys for supported chat models, but that sends requests to a remote provider and does not reveal the model’s training corpus. Privacy Mode governs Cursor handling; it cannot establish the provider’s rights practices. Select a model/provider with a clear training-data and opt-out policy before using private code.", "argument": "Cursor lets users bring provider API keys for supported chat models, but that sends requests to a remote provider and does not reveal the model’s training corpus. Privacy Mode governs Cursor handling; it cannot establish the provider’s rights practices. Select a model/provider with a clear training-data and opt-out policy before using private code.", "assessmentType": "conditional", "score": null, "href": "/harnesses/cursor/#training-rights" }, "privacy": { "summary": "Cursor’s Privacy Mode and its cloud background-agent path handle code differently.", "sourceUrls": [ "https://docs.cursor.com/account/privacy", "https://docs.cursor.com/background-agent" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 5, "scoreReason": "Privacy Mode offers controls for standard requests, but background agents use cloud VMs and can retain task data separately. Check the mode and provider for each workflow; BYOK alone does not mean the editor never receives the content.", "argument": "Privacy Mode offers controls for standard requests, but background agents use cloud VMs and can retain task data separately. Check the mode and provider for each workflow; BYOK alone does not mean the editor never receives the content.", "assessmentType": "evidence", "score": 5, "href": "/harnesses/cursor/#privacy" }, "openness": { "summary": "Published model choice does not make the closed commercial editor open.", "sourceUrls": [ "https://docs.cursor.com/account/privacy", "https://docs.cursor.com/models" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 2, "scoreReason": "Cursor’s documentation describes a commercial editor and model choices but does not publish the editor under an open-source license. Choosing a different model or API route changes inference, not whether the editor implementation can be inspected. Assess the client separately from its providers.", "argument": "Cursor’s documentation describes a commercial editor and model choices but does not publish the editor under an open-source license. Choosing a different model or API route changes inference, not whether the editor implementation can be inspected. Assess the client separately from its providers.", "assessmentType": "evidence", "score": 2, "href": "/harnesses/cursor/#openness" }, "labour": { "summary": "Anysphere’s product privacy documentation does not describe its employees, suppliers, or model-data workers in enough detail to assess labor conditions.", "sourceUrls": [ "https://docs.cursor.com/account/privacy", "https://cursor.com/" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "The privacy and product pages reviewed for Cursor do not describe Anysphere’s supplier workforce or data-labeling conditions. That leaves a real information gap, not proof that workers are treated poorly. Ask Anysphere which model providers and data-work vendors support each feature and what independent worker protections apply.", "argument": "The privacy and product pages reviewed for Cursor do not describe Anysphere’s supplier workforce or data-labeling conditions. That leaves a real information gap, not proof that workers are treated poorly. Ask Anysphere which model providers and data-work vendors support each feature and what independent worker protections apply.", "assessmentType": "argument", "score": null, "href": "/harnesses/cursor/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://docs.cursor.com/settings/api-keys" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "The reviewed sources do not name Cursor in a defense or surveillance contract. Cursor can call several model providers, so a service relationship at OpenAI, Google, or Anthropic would belong to that provider unless Cursor’s own deployment is shown. Check the provider and any enterprise integrations actually enabled; silence in this review is not a clean record.", "argument": "The reviewed sources do not name Cursor in a defense or surveillance contract. Cursor can call several model providers, so a service relationship at OpenAI, Google, or Anthropic would belong to that provider unless Cursor’s own deployment is shown. Check the provider and any enterprise integrations actually enabled; silence in this review is not a clean record.", "assessmentType": "conditional", "score": null, "href": "/harnesses/cursor/#military-contracts" }, "israel-palestine": { "summary": "We did not find Anysphere among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Anysphere is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/cursor/#israel-palestine" } }, "sources": [ { "url": "https://docs.cursor.com/account/privacy", "title": "Cursor privacy settings and request handling", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.cursor.com/models", "title": "Cursor model provider choices", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.cursor.com/background-agent", "title": "Official product documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.cursor.com/settings/api-keys", "title": "Cursor provider API-key options", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cursor.com/", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/harnesses/cursor/", "overall": { "score": 3.5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "red", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 16 } }, { "id": "github-copilot", "name": "GitHub Copilot", "category": "harnesses", "organization": "GitHub / Microsoft", "website": "https://github.com/features/copilot", "summary": "Commercial coding assistant embedded in IDEs, GitHub, and terminal workflows; models and hosting paths vary.", "scope": "This covers product controls, not the entire Microsoft cloud stack or underlying models.", "dimensions": { "energy": { "summary": "GitHub Copilot does not publish service-specific energy or water data; the model and serving provider determine most inference impact.", "sourceUrls": [ "https://docs.github.com/en/copilot/how-tos/use-ai-models/change-the-chat-model?tool=vscode" ], "confidence": "low", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Copilot lets users switch among supported chat models, but inline suggestions and some clients may use different models. The documentation offers choice, not energy or water accounting. Pick a smaller capable model where available and request model/region-level figures from GitHub before making a low-impact claim.", "argument": "Copilot lets users switch among supported chat models, but inline suggestions and some clients may use different models. The documentation offers choice, not energy or water accounting. Pick a smaller capable model where available and request model/region-level figures from GitHub before making a low-impact claim.", "assessmentType": "argument", "score": null, "href": "/harnesses/github-copilot/#energy" }, "training-rights": { "summary": "GitHub Copilot is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://docs.github.com/en/copilot/reference/ai-models/model-hosting", "https://docs.github.com/en/copilot/concepts/completions/code-referencing" ], "confidence": "low", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "Copilot lets users select among models on supported plans, but model-specific hosting and training terms differ. That choice is useful control, not evidence that the models’ training corpora were licensed or that code-referencing prevents training. Check the selected model’s hosting and terms, and prefer a provider that publishes rights provenance.", "argument": "Copilot lets users select among models on supported plans, but model-specific hosting and training terms differ. That choice is useful control, not evidence that the models’ training corpora were licensed or that code-referencing prevents training. Check the selected model’s hosting and terms, and prefer a provider that publishes rights provenance.", "assessmentType": "conditional", "score": null, "href": "/harnesses/github-copilot/#training-rights" }, "privacy": { "summary": "Copilot data handling varies by subscription, selected model, and hosting provider.", "sourceUrls": [ "https://docs.github.com/en/copilot/reference/ai-models/model-hosting" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 6, "scoreReason": "The published rules distinguish individual from Business/Enterprise plans and document different provider paths. Check the exact plan and model; do not generalize an enterprise exclusion to a personal account or every model host.", "argument": "The published rules distinguish individual from Business/Enterprise plans and document different provider paths. Check the exact plan and model; do not generalize an enterprise exclusion to a personal account or every model host.", "assessmentType": "evidence", "score": 6, "href": "/harnesses/github-copilot/#privacy" }, "openness": { "summary": "The product is proprietary; model choice and code-reference features do not expose its implementation.", "sourceUrls": [ "https://docs.github.com/en/copilot/reference/ai-models/model-hosting", "https://docs.github.com/en/copilot/concepts/completions/code-referencing" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 2, "scoreReason": "GitHub documents Copilot as a hosted commercial product and explains its model and code-reference features, but it does not publish the product implementation under an open-source license. Model selection and code-reference controls do not make the service code inspectable. The editor integration and selected model are separate layers.", "argument": "GitHub documents Copilot as a hosted commercial product and explains its model and code-reference features, but it does not publish the product implementation under an open-source license. Model selection and code-reference controls do not make the service code inspectable. The editor integration and selected model are separate layers.", "assessmentType": "evidence", "score": 2, "href": "/harnesses/github-copilot/#openness" }, "labour": { "summary": "Company context: Microsoft, GitHub’s parent company. Microsoft’s FY2025 Human Rights Transparency Report describes supplier monitoring and reports findings in its Devices/Cloud Supply Chain and Provisioning scope, including 34 forced-labor/trafficking indicators (33 closed, one corrective plan), 50 fair-compensation findings (46 closed, four plans), and 75 working-hours/rest findings (68 closed, seven plans). These are supplier-audit findings, not AI-annotation-specific results. This is a record of the company behind GitHub Copilot; it does not establish use of this particular product or release.", "sourceUrls": [ "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Human-Rights-Transparency-Report-2025.pdf", "https://www.microsoft.com/en-us/procurement/supplier-conduct", "https://github.com/features/copilot" ], "confidence": "high", "unknowns": "Working conditions for the people building or supporting GitHub Copilot are not established by this company-wide record. The published findings cover specified supplier audit scopes, not all suppliers, model evaluation labor, or all remediation outcomes.", "suggestedScore": 6, "scoreReason": "Company-level assessment: Microsoft reports supplier findings and closure plans, but the reported scope is not specific to AI annotation and some plans remain open. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "Microsoft reports supplier findings and closure plans, but the reported scope is not specific to AI annotation and some plans remain open. This is relevant when deciding whose business to support, while pay and working conditions for the team behind GitHub Copilot remain unestablished. Ask for current worker outcomes for that team and its suppliers.", "assessmentScope": "Company record", "productArgument": "Microsoft’s Supplier Code sets expectations for subcontractors and describes compliance processes, but it is not an audit of the specific models offered through Copilot. It does not name their labeling vendors, pay, or safety provisions. Ask Microsoft/GitHub for model-specific workforce evidence before treating a supplier policy as an outcome.", "score": 6, "href": "/harnesses/github-copilot/#labour" }, "military-contracts": { "summary": "Company context: Microsoft, GitHub’s parent company. Microsoft’s final 2026 review says its external investigation found evidence supporting elements of 2025 reporting, including IMOD Azure storage consumption in the Netherlands and AI-service use. Microsoft says it disabled specified storage/AI subscriptions in September 2025, while continuing other software, professional services, Azure and Azure AI services to IMOD. This is a record of the company behind GitHub Copilot; it does not establish use of this particular product or release.", "sourceUrls": [ "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Summary-of-2025-External-Investigation-Follow-Up.pdf", "https://learn.microsoft.com/en-us/compliance/us-government/gov-jwcc" ], "confidence": "high", "unknowns": "Deployment of GitHub Copilot under these contracts is not established. The report does not access IMOD customer content, does not establish the content/use of every service, and does not say the wider commercial relationship ended.", "suggestedScore": 2, "scoreReason": "Company-level assessment: The company confirms defense-ministry Azure/AI use and a review supporting elements of serious surveillance-related reporting; it stopped specified services but continues others. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "The company confirms defense-ministry Azure/AI use and a review supporting elements of serious surveillance-related reporting; it stopped specified services but continues others. This is relevant when deciding whose business to support. Deployment of GitHub Copilot or this particular model release under those contracts remains unestablished; consider the company relationship separately from the product’s actual use.", "assessmentScope": "Company record", "productArgument": "Microsoft is a JWCC cloud provider in the Department of Defense procurement. This establishes a Microsoft-level defense relationship but not GitHub Copilot use in JWCC workloads. Buyers can weigh Microsoft’s broader contract portfolio while keeping it separate from product deployment.", "score": 2, "href": "/harnesses/github-copilot/#military-contracts" }, "israel-palestine": { "summary": "Microsoft is named as a consumer boycott priority target in the BNC guide. GitHub is part of Microsoft; the classification follows its parent company.", "argument": "If following this BNC campaign is a requirement, consider alternatives to paid Microsoft products where practical. Check the model and host this tool connects to as well as the company providing the tool. This label follows the named company’s campaign status; it does not claim that BDS separately reviewed this exact product.", "assessmentType": "evidence", "confidence": "high", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://bdsmovement.net/microsoft", "https://docs.github.com/en/site-policy/privacy-policies/github-general-privacy-statement", "https://blogs.microsoft.com/blog/2018/10/26/microsoft-completes-github-acquisition/" ], "bdsStatus": "priority-boycott", "statusLabel": "Boycott target", "reviewedAt": "2026-10-23", "assessmentScope": "Parent company", "bdsMatch": { "target": "Microsoft", "status": "priority-boycott", "scope": "Parent company", "relationship": "GitHub is part of Microsoft; the classification follows its parent company.", "application": "If following this BNC campaign is a requirement, consider alternatives to paid Microsoft products where practical.", "sourceIds": [ "microsoft", "github", "github-parent" ] }, "upstreamExposure": null, "businessContext": { "summary": "Company context: Microsoft, GitHub’s parent company. Microsoft’s final 2026 review says Covington found evidence supporting elements of reporting about IMOD Azure storage and AI-service use, including Azure storage consumption in the Netherlands. Microsoft says it disabled specified subscriptions/services in September 2025 but continues providing IMOD with other Azure, Azure AI, software, and professional services. The report did not access customer content and does not establish all downstream uses. This is a record of the company behind GitHub Copilot; it does not establish use of this particular product or release.", "unknowns": "How GitHub Copilot is used in this relationship is not established. Microsoft describes only selected service suspensions; the report does not establish all IMOD use, civilian impact, or the full scope of continuing services.", "sourceUrls": [ "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Summary-of-2025-External-Investigation-Follow-Up.pdf" ] }, "score": null, "href": "/harnesses/github-copilot/#israel-palestine" } }, "sources": [ { "url": "https://docs.github.com/en/copilot/reference/ai-models/model-hosting", "title": "GitHub Copilot model hosting and data use", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.github.com/en/copilot/concepts/completions/code-referencing", "title": "GitHub Copilot code referencing controls", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://learn.microsoft.com/en-us/compliance/us-government/gov-jwcc", "title": "Microsoft Joint Warfighting Cloud Capability program", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.github.com/en/copilot/how-tos/use-ai-models/change-the-chat-model?tool=vscode", "title": "GitHub Copilot model selection", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.microsoft.com/en-us/procurement/supplier-conduct", "title": "Microsoft Supplier Code of Conduct", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/features/copilot", "title": "Copilot", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Human-Rights-Transparency-Report-2025.pdf", "title": "Microsoft Human Rights Transparency Report 2025", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Summary-of-2025-External-Investigation-Follow-Up.pdf", "title": "Microsoft Summary of 2025 External Investigation and Follow Up", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/microsoft", "title": "BNC: Microsoft campaign and subsidiaries guidance", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://docs.github.com/en/site-policy/privacy-policies/github-general-privacy-statement", "title": "GitHub privacy statement: Microsoft affiliation, effective April 2026", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://blogs.microsoft.com/blog/2018/10/26/microsoft-completes-github-acquisition/", "title": "Microsoft: completed acquisition of GitHub", "kind": "primary", "checkedAt": "2026-10-23" } ], "href": "/harnesses/github-copilot/", "overall": { "score": 4, "rated": 4, "total": 6, "coverage": 0.6666666666666666, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 14 } }, { "id": "gemini-cli", "name": "Gemini CLI", "category": "harnesses", "organization": "Google", "website": "https://github.com/google-gemini/gemini-cli", "summary": "Open-source command-line coding agent connecting a terminal workspace to Google Gemini services or supported API backends.", "scope": "Privacy terms vary with Google-account, Gemini API, or Vertex AI authentication; Google contracts do not establish this CLI deployment.", "dimensions": { "energy": { "summary": "Gemini CLI does not publish service-specific energy or water data; the model and serving provider determine most inference impact.", "sourceUrls": [ "https://github.com/google-gemini/gemini-cli/blob/main/docs/resources/tos-privacy.md" ], "confidence": "low", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Gemini CLI routes requests through Code Assist, Gemini API, or Vertex AI according to the login method. A Vertex region may offer a clearer infrastructure boundary, but the CLI’s terms page gives no energy data for the service. Select a region and model with substantiated emissions reporting if that is part of your purchasing test.", "argument": "Gemini CLI routes requests through Code Assist, Gemini API, or Vertex AI according to the login method. A Vertex region may offer a clearer infrastructure boundary, but the CLI’s terms page gives no energy data for the service. Select a region and model with substantiated emissions reporting if that is part of your purchasing test.", "assessmentType": "argument", "score": null, "href": "/harnesses/gemini-cli/#energy" }, "training-rights": { "summary": "Gemini CLI is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://github.com/google-gemini/gemini-cli/blob/main/docs/resources/tos-privacy.md" ], "confidence": "low", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "Gemini CLI routes through Google Code Assist, Gemini API, or Vertex AI depending on authentication. Those paths have different terms, but none of the cited CLI documentation is a complete source-level rights ledger for Gemini models. Pick a paid/enterprise path with suitable data controls and ask Google for model-specific provenance if permission is a hard requirement.", "argument": "Gemini CLI routes through Google Code Assist, Gemini API, or Vertex AI depending on authentication. Those paths have different terms, but none of the cited CLI documentation is a complete source-level rights ledger for Gemini models. Pick a paid/enterprise path with suitable data controls and ask Google for model-specific provenance if permission is a hard requirement.", "assessmentType": "conditional", "score": null, "href": "/harnesses/gemini-cli/#training-rights" }, "privacy": { "summary": "Gemini CLI applies the privacy terms of the chosen sign-in path: Code Assist, Gemini API, or Vertex AI.", "sourceUrls": [ "https://github.com/google-gemini/gemini-cli/blob/main/docs/resources/tos-privacy.md" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 5, "scoreReason": "The CLI is the same front end but the data rules change with authentication and paid/unpaid service. Choose enterprise Vertex/Code Assist only if its terms fit your needs, and review the consumer API policy separately.", "argument": "The CLI is the same front end but the data rules change with authentication and paid/unpaid service. Choose enterprise Vertex/Code Assist only if its terms fit your needs, and review the consumer API policy separately.", "assessmentType": "evidence", "score": 5, "href": "/harnesses/gemini-cli/#privacy" }, "openness": { "summary": "Gemini CLI software is Apache-2.0; the backend model and service terms are separate.", "sourceUrls": [ "https://github.com/google-gemini/gemini-cli/blob/main/docs/resources/tos-privacy.md", "https://github.com/google-gemini/gemini-cli/blob/main/docs/changelogs/latest.md" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 9, "scoreReason": "The Gemini CLI repository is Apache-2.0, so its client code can be inspected and modified. The Gemini model and hosted services remain governed by separate terms. Open client code does not establish access to model weights or training data.", "argument": "The Gemini CLI repository is Apache-2.0, so its client code can be inspected and modified. The Gemini model and hosted services remain governed by separate terms. Open client code does not establish access to model weights or training data.", "assessmentType": "evidence", "score": 9, "href": "/harnesses/gemini-cli/#openness" }, "labour": { "summary": "Company context: Google. Google’s supplier code applies labor and human-rights requirements across direct, temporary, migrant, contract, and subcontracted workers; it allows audits, requires remediation, and sets a 60-hour weekly limit with stated exceptions. This is a supplier policy, not an audit of AI data workers. This is a record of the company behind Gemini CLI; it does not establish use of this particular product or release.", "sourceUrls": [ "https://about.google/company-info/supplier-code-of-conduct/", "https://github.com/google-gemini/gemini-cli" ], "confidence": "high", "unknowns": "Working conditions for the people building or supporting Gemini CLI are not established by this company-wide record. No current AI-annotation-specific outcomes, public audit findings, or remediation data for Vertex-related data work were established.", "suggestedScore": 4, "scoreReason": "Company-level assessment: Google publishes broad supplier protections and audit rights, but this review found no worker-level results for AI data work. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "Google publishes broad supplier protections and audit rights, but this review found no worker-level results for AI data work. This is relevant when deciding whose business to support, while pay and working conditions for the team behind Gemini CLI remain unestablished. Ask for current worker outcomes for that team and its suppliers.", "assessmentScope": "Company record", "productArgument": "Google’s Supplier Code covers vendors and subcontractors and describes audit rights, but does not report conditions for workers who prepared the models behind Gemini CLI. The policy is a baseline expectation rather than proof of current worker outcomes. Ask Google for model-specific vendors, wages, safety support, and audit findings.", "score": 4, "href": "/harnesses/gemini-cli/#labour" }, "military-contracts": { "summary": "Company context: Google. Google documents access for Department of War personnel and classified defense workflows. Separately, Israel’s procurement administration identifies Google as a Nimbus public-cloud provider; AFSC and Who Profits describe military/security use, while the Guardian reports contract terms limiting service withdrawal. These sources do not establish Google’s role in a particular attack. This is a record of the company behind Gemini CLI; it does not establish use of this particular product or release.", "sourceUrls": [ "https://cloud.google.com/blog/topics/public-sector/announcements-and-highlights-from-google-public-sector-summit-2026", "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/alphabet", "https://cloud.google.com/blog/topics/public-sector/google-public-sector-awarded-200-million-contract-to-accelerate-ai-and-cloud-capabilities-across-department-of-defenses-chief-digital-and-artificial-intelligence-office-cdao" ], "confidence": "high", "unknowns": "Deployment of Gemini CLI under these contracts is not established. The source does not disclose contract terms or evaluate downstream deployment outcomes.", "suggestedScore": 3, "scoreReason": "Company-level assessment: Google documents U.S. Department of War and classified defense workflows; Israel’s procurement record identifies Google as a Nimbus cloud provider, and independent research/reporting describes military or security use and limits on withdrawal. The reviewed evidence does not attribute a specific harmful operation to Google. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "conditional", "argument": "Google documents U.S. Department of War and classified defense workflows; Israel’s procurement record identifies Google as a Nimbus cloud provider, and independent research/reporting describes military or security use and limits on withdrawal. The reviewed evidence does not attribute a specific harmful operation to Google. This is relevant when deciding whose business to support. Deployment of Gemini CLI or this particular model release under those contracts remains unestablished; consider the company relationship separately from the product’s actual use.", "assessmentScope": "Company record", "productArgument": "Google Public Sector announced a DoD CDAO contract with a $200 million ceiling. The announcement establishes a parent-company defense relationship but does not identify Gemini CLI or its model service as the deployed tool. Buyers can ask which Google services and workloads the contract covers.", "score": 3, "href": "/harnesses/gemini-cli/#military-contracts" }, "israel-palestine": { "summary": "Google / Alphabet is named as a pressure target in the BNC’s official campaign sources. The product is developed or supplied by Google / Alphabet.", "argument": "If following this BNC campaign is a requirement, consider alternatives to paid Google / Alphabet products where practical. Check the model and host this tool connects to as well as the company providing the tool. This label follows the named company’s campaign status; it does not claim that BDS separately reviewed this exact product.", "assessmentType": "evidence", "confidence": "high", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "pressure", "statusLabel": "Pressure target", "reviewedAt": "2026-10-23", "assessmentScope": "Developer / service company", "bdsMatch": { "target": "Google / Alphabet", "status": "pressure", "scope": "Developer / service company", "relationship": "The product is developed or supplied by Google / Alphabet.", "application": "If following this BNC campaign is a requirement, consider alternatives to paid Google / Alphabet products where practical.", "sourceIds": [] }, "upstreamExposure": null, "businessContext": { "summary": "Company context: Google. Israel’s procurement administration says Google was selected with AWS for Nimbus public-cloud services and lists the Ministry of Defense and IDF among project partners. The Guardian’s investigation of leaked agreement documents reports that Nimbus terms limited Google’s ability to suspend service, including after alleged policy violations. Who Profits and AFSC describe cloud and AI use by military/security bodies, including surveillance-related capabilities; these are attributed research claims, not proof of Google’s role in a specific attack. Both companies denied circumventing legal obligations in response to the Guardian investigation. This is a record of the company behind Gemini CLI; it does not establish use of this particular product or release.", "unknowns": "How Gemini CLI is used in this relationship is not established. The government record establishes the contract and project participants; the investigative reporting describes contract controls, and research organizations make broader use claims. The reviewed evidence does not establish Google’s contribution to a particular operation, injury, or other specific harm.", "sourceUrls": [ "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/alphabet", "https://cloud.google.com/blog/topics/inside-google-cloud/google-cloud-selected-to-provide-cloud-services-to-the-state-of-israel" ] }, "score": null, "href": "/harnesses/gemini-cli/#israel-palestine" } }, "sources": [ { "url": "https://github.com/google-gemini/gemini-cli/blob/main/docs/resources/tos-privacy.md", "title": "Gemini CLI service-specific terms and privacy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/google-gemini/gemini-cli/blob/main/docs/changelogs/latest.md", "title": "Official license or documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cloud.google.com/blog/topics/inside-google-cloud/google-cloud-selected-to-provide-cloud-services-to-the-state-of-israel", "title": "Google Cloud state of Israel announcement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://about.google/company-info/supplier-code-of-conduct/", "title": "Google Supplier Code of Conduct", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/google-gemini/gemini-cli", "title": "Gemini Cli", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cloud.google.com/blog/topics/public-sector/google-public-sector-awarded-200-million-contract-to-accelerate-ai-and-cloud-capabilities-across-department-of-defenses-chief-digital-and-artificial-intelligence-office-cdao", "title": "Google Public Sector DoD CDAO agreement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cloud.google.com/blog/topics/public-sector/announcements-and-highlights-from-google-public-sector-summit-2026", "title": "Announcements and highlights from Google Public Sector Summit 2026", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://mr.gov.il/ilgstorefront/en/news/details/111222", "title": "Nimbus Tender — Israel Government Procurement Administration", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "title": "Revealed: Israel demanded Google and Amazon use secret “wink” to sidestep legal orders", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://www.whoprofits.org/publications/report/164", "title": "Digitizing Occupation: The Role of Big Tech in Israeli Military Infrastructure", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://investigate.afsc.org/company/alphabet", "title": "Alphabet / Google — AFSC Investigate", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/harnesses/gemini-cli/", "overall": { "score": 5.3, "rated": 4, "total": 6, "coverage": 0.6666666666666666, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 10 } }, { "id": "windsurf", "name": "Devin Desktop (formerly Windsurf)", "category": "harnesses", "organization": "Cognition", "website": "https://windsurf.com/", "summary": "Commercial coding IDE and agent workspace, now branded Devin Desktop; it combines the Windsurf editor with local and cloud agents.", "scope": "This profile follows the Windsurf IDE into Devin Desktop, which the company identifies as Windsurf’s new name. It covers the current commercial editor and agent service, not a legacy Codeium extension or unrelated source repository. The current master services agreement grants limited access and reserves rights in the service; model and hosting choices remain separate.", "dimensions": { "energy": { "summary": "Windsurf does not publish service-specific energy or water data; the model and serving provider determine most inference impact.", "sourceUrls": [ "https://windsurf.com/MSA" ], "confidence": "low", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "The current product page advertises both local and cloud agents, but the reviewed materials provide no per-task electricity, water or emissions data. Local execution moves some energy use to the user’s machine; it does not establish a lower footprint. Measure the actual local setup or ask Cognition for comparable figures for the selected cloud model and region.", "argument": "The current product page advertises both local and cloud agents, but the reviewed materials provide no per-task electricity, water or emissions data. Local execution moves some energy use to the user’s machine; it does not establish a lower footprint. Measure the actual local setup or ask Cognition for comparable figures for the selected cloud model and region.", "assessmentType": "argument", "score": null, "href": "/harnesses/windsurf/#energy" }, "training-rights": { "summary": "Windsurf is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://windsurf.com/MSA" ], "confidence": "low", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "The current agreement covers customer data sent through the service, while the product page describes access to models and agents; neither provides a training-data provenance ledger for each model. A promise about submitted prompts is different from permission for material used to train the model. Check the selected model’s provider and terms separately.", "argument": "The current agreement covers customer data sent through the service, while the product page describes access to models and agents; neither provides a training-data provenance ledger for each model. A promise about submitted prompts is different from permission for material used to train the model. Check the selected model’s provider and terms separately.", "assessmentType": "conditional", "score": null, "href": "/harnesses/windsurf/#training-rights" }, "privacy": { "summary": "Windsurf’s agreement promises no training on customer inputs and deletion after generation, with exceptions for enabled persistence features and integrations.", "sourceUrls": [ "https://windsurf.com/MSA", "https://windsurf.com/docs/DPA.pdf" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 7, "scoreReason": "The agreement sets protections for customer data, while the data-processing terms and enabled integrations determine additional processing paths. Those terms do not make every plan or feature identical. Before sending sensitive code, check the current account terms, selected agent, integrations and persistence settings.", "argument": "The agreement sets protections for customer data, while the data-processing terms and enabled integrations determine additional processing paths. Those terms do not make every plan or feature identical. Before sending sensitive code, check the current account terms, selected agent, integrations and persistence settings.", "assessmentType": "evidence", "score": 7, "href": "/harnesses/windsurf/#privacy" }, "openness": { "summary": "Commercial proprietary product; no public code license establishes an open implementation.", "sourceUrls": [ "https://windsurf.com/MSA", "https://windsurf.com/docs/DPA.pdf" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 2, "scoreReason": "Cognition’s agreement grants limited access to the service and reserves rights in its software and related intellectual property. That makes the current editor a commercial proprietary product; choosing an open model or installing an open extension does not open the IDE implementation. This profile follows the current Devin Desktop product, not a different Codeium repository.", "argument": "Cognition’s agreement grants limited access to the service and reserves rights in its software and related intellectual property. That makes the current editor a commercial proprietary product; choosing an open model or installing an open extension does not open the IDE implementation. This profile follows the current Devin Desktop product, not a different Codeium repository.", "assessmentType": "evidence", "score": 2, "href": "/harnesses/windsurf/#openness" }, "labour": { "summary": "Cognition discloses subprocessors for service processing, but that is not a report on their labor conditions or on model-data workers.", "sourceUrls": [ "https://cognition.ai/Cognition-DPA.pdf", "https://windsurf.com/editor" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "Cognition’s data-processing agreement names subprocessors involved in service processing, but it does not describe their workers’ pay or conditions or the labor behind model training. A vendor list helps trace who may process data; it is not a labor audit. Ask Cognition and the selected model provider for worker standards and independent checks.", "argument": "Cognition’s data-processing agreement names subprocessors involved in service processing, but it does not describe their workers’ pay or conditions or the labor behind model training. A vendor list helps trace who may process data; it is not a labor audit. Ask Cognition and the selected model provider for worker standards and independent checks.", "assessmentType": "argument", "score": null, "href": "/harnesses/windsurf/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://windsurf.com/MSA" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "The reviewed product and service agreements do not identify this IDE in a defense workload. Cognition’s current product includes local and cloud agents, so host and model-provider contracts need separate review. Ask which services apply to your plan before drawing a conclusion about a specific deployment.", "argument": "The reviewed product and service agreements do not identify this IDE in a defense workload. Cognition’s current product includes local and cloud agents, so host and model-provider contracts need separate review. Ask which services apply to your plan before drawing a conclusion about a specific deployment.", "assessmentType": "conditional", "score": null, "href": "/harnesses/windsurf/#military-contracts" }, "israel-palestine": { "summary": "We did not find Cognition among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Cognition is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/windsurf/#israel-palestine" } }, "sources": [ { "url": "https://windsurf.com/MSA", "title": "Windsurf master services agreement and customer data terms", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://windsurf.com/docs/DPA.pdf", "title": "Official privacy or data-processing disclosure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cognition.ai/Cognition-DPA.pdf", "title": "Cognition Data Processing Agreement and subprocessors", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://windsurf.com/editor", "title": "Official product page: Devin Desktop is the new name for Windsurf and retains its IDE", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "releaseNote": "Product-name note: Windsurf’s official product page says Devin Desktop is the new name for Windsurf and that the IDE, extensions, workflows and settings continue through the rename.", "href": "/harnesses/windsurf/", "overall": { "score": 4.5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 13 } }, { "id": "cline", "name": "Cline", "category": "harnesses", "organization": "Cline", "website": "https://cline.bot/", "summary": "Open-source agentic coding extension and related CLI/SDK configurable with model providers.", "scope": "Prompts and code flow to configured providers. Official privacy and telemetry descriptions conflict across dated pages.", "dimensions": { "energy": { "summary": "Cline can route inference to a local model; that gives the user control over the serving machine, not a verified reduction in energy or water.", "sourceUrls": [ "https://cline.bot/" ], "confidence": "medium", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Cline can send requests to hosted providers or run models through Ollama/LM Studio on your own hardware. Local serving gives you a place to measure electricity, but large models may use substantial device power and the product publishes no comparative results. Measure a bounded task on your hardware and compare it with a named hosted route before claiming a reduction.", "argument": "Cline can send requests to hosted providers or run models through Ollama/LM Studio on your own hardware. Local serving gives you a place to measure electricity, but large models may use substantial device power and the product publishes no comparative results. Measure a bounded task on your hardware and compare it with a named hosted route before claiming a reduction.", "assessmentType": "conditional", "score": null, "href": "/harnesses/cline/#energy" }, "training-rights": { "summary": "Cline’s current terms say Cline-routed provider content may be used for model training unless the provider offers an opt-out and it is enabled; own-key and local paths differ.", "sourceUrls": [ "https://cline.bot/tos", "https://cline.bot/" ], "confidence": "high", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": 4, "scoreReason": "Cline supports local weights and a broad set of hosted providers. Its current terms make clear that provider data may be used for training unless the user opts out where available, so model selection alone does not guarantee rights. Choose a documented local model or verify the selected provider’s opt-out and corpus disclosures before sharing code.", "argument": "Cline supports local weights and a broad set of hosted providers. Its current terms make clear that provider data may be used for training unless the user opts out where available, so model selection alone does not guarantee rights. Choose a documented local model or verify the selected provider’s opt-out and corpus disclosures before sharing code.", "assessmentType": "evidence", "score": 4, "href": "/harnesses/cline/#training-rights" }, "privacy": { "summary": "Cline’s current terms say telemetry is on by default, distinguish Cline-provided keys from personal keys, and assign provider retention to downstream providers.", "sourceUrls": [ "https://cline.bot/privacy", "https://cline.bot/tos", "https://cline.bot/blog/introducing-anonymous-telemetry-in-cline" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 4, "scoreReason": "Current Terms of Service (September 25, 2025) say telemetry is on by default and may be disabled; they also permit service-improvement use of user content outside per-seat subscriptions and say providers may train unless opted out. The 2025 blog says telemetry is opt-in, so the newer terms are the safer source for the current default; use your own key/local model and disable telemetry if you need tighter control.", "argument": "Current Terms of Service (September 25, 2025) say telemetry is on by default and may be disabled; they also permit service-improvement use of user content outside per-seat subscriptions and say providers may train unless opted out. The 2025 blog says telemetry is opt-in, so the newer terms are the safer source for the current default; use your own key/local model and disable telemetry if you need tighter control.", "assessmentType": "evidence", "score": 4, "href": "/harnesses/cline/#privacy" }, "openness": { "summary": "Cline public repository is Apache-2.0; provider models remain separate.", "sourceUrls": [ "https://github.com/cline/cline", "https://cline.bot/privacy" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 9, "scoreReason": "Cline’s Apache-2.0 repository allows users to inspect, modify and redistribute the client. Its connected models and provider services have their own licenses and terms. The open extension does not make a hosted model open.", "argument": "Cline’s Apache-2.0 repository allows users to inspect, modify and redistribute the client. Its connected models and provider services have their own licenses and terms. The open extension does not make a hosted model open.", "assessmentType": "evidence", "score": 9, "href": "/harnesses/cline/#openness" }, "labour": { "summary": "Cline’s public repository makes code contributions inspectable; it does not establish whether contributors are paid or how Cline’s model providers treat data workers.", "sourceUrls": [ "https://github.com/cline/cline" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "Cline’s repository makes its code and contribution history visible, but it does not disclose whether maintainers are paid or how model providers treat data workers. Open code is not the same as fair work. Consider supporting maintainers and ask the chosen provider for its current data-worker disclosures.", "argument": "Cline’s repository makes its code and contribution history visible, but it does not disclose whether maintainers are paid or how model providers treat data workers. Open code is not the same as fair work. Consider supporting maintainers and ask the chosen provider for its current data-worker disclosures.", "assessmentType": "conditional", "score": null, "href": "/harnesses/cline/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://cline.bot/" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "Cline’s current product pages emphasize user-selected and local providers; the sources reviewed do not name a Cline defense deployment. A model provider’s contract should not be attributed to Cline without evidence that the product was used. For sensitive development, choose a local provider or ask Cline and the selected host about contract restrictions.", "argument": "Cline’s current product pages emphasize user-selected and local providers; the sources reviewed do not name a Cline defense deployment. A model provider’s contract should not be attributed to Cline without evidence that the product was used. For sensitive development, choose a local provider or ask Cline and the selected host about contract restrictions.", "assessmentType": "conditional", "score": null, "href": "/harnesses/cline/#military-contracts" }, "israel-palestine": { "summary": "We did not find Cline among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Cline is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/cline/#israel-palestine" } }, "sources": [ { "url": "https://github.com/cline/cline", "title": "Official source repository or product page", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cline.bot/privacy", "title": "Official privacy or data-processing disclosure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cline.bot/tos", "title": "Cline Terms of Service (last modified 2025-09-25)", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cline.bot/blog/introducing-anonymous-telemetry-in-cline", "title": "Official telemetry documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cline.bot/", "title": "Cline current product, local-provider and bring-your-own-key options", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/harnesses/cline/", "overall": { "score": 5.7, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 9 } }, { "id": "roo-code", "name": "Roo Code (archived)", "category": "harnesses", "organization": "Roo Code Inc.", "website": "https://github.com/RooCodeInc/Roo-Code", "summary": "Former open-source coding-agent extension. Project announced shutdown; repository archived, so this is a historical profile.", "scope": "Privacy and telemetry evidence is historical and does not describe a currently maintained service.", "dimensions": { "energy": { "summary": "Roo Code (archived) can route inference to a local model; that gives the user control over the serving machine, not a verified reduction in energy or water.", "sourceUrls": [ "https://roocodeinc.github.io/Roo-Code/providers/ollama/" ], "confidence": "medium", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Roo Code’s former extension supported Ollama and LM Studio, so a maintained fork could run inference on local hardware. That can make electricity observable at the device, but does not prove the local route is more efficient than cloud inference. Since the project is archived, first verify the model server and fork still work, then measure energy per completed task.", "argument": "Roo Code’s former extension supported Ollama and LM Studio, so a maintained fork could run inference on local hardware. That can make electricity observable at the device, but does not prove the local route is more efficient than cloud inference. Since the project is archived, first verify the model server and fork still work, then measure energy per completed task.", "assessmentType": "conditional", "score": null, "href": "/harnesses/roo-code/#energy" }, "training-rights": { "summary": "Roo Code (archived) is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://github.com/RooCodeInc/Roo-Code/blob/main/PRIVACY.md", "https://roocodeinc.github.io/Roo-Code/providers/ollama/" ], "confidence": "medium", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "Roo Code’s archived extension supported local models and many remote providers. The local path offers more control over the selected model, but the archived privacy notice does not establish any model’s training permissions. If maintaining a fork, select a model with published data provenance and do not rely on the old provider list as a rights review.", "argument": "Roo Code’s archived extension supported local models and many remote providers. The local path offers more control over the selected model, but the archived privacy notice does not establish any model’s training permissions. If maintaining a fork, select a model with published data provenance and do not rely on the old provider list as a rights review.", "assessmentType": "conditional", "score": null, "href": "/harnesses/roo-code/#training-rights" }, "privacy": { "summary": "Roo Code’s historical privacy disclosure describes provider routing and telemetry, but the extension is archived.", "sourceUrls": [ "https://github.com/RooCodeInc/Roo-Code/blob/main/PRIVACY.md" ], "confidence": "low", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": null, "scoreReason": "Roo Code supported local models, which could keep inference on the user’s machine, but the archived extension has no current maintenance assurance. Treat its privacy and security notes as historical and avoid assuming the former controls still receive fixes.", "argument": "Roo Code supported local models, which could keep inference on the user’s machine, but the archived extension has no current maintenance assurance. Treat its privacy and security notes as historical and avoid assuming the former controls still receive fixes.", "assessmentType": "evidence", "score": null, "href": "/harnesses/roo-code/#privacy" }, "openness": { "summary": "Archived extension code is Apache-2.0, but the project is no longer actively maintained.", "sourceUrls": [ "https://github.com/RooCodeInc/Roo-Code", "https://github.com/RooCodeInc/Roo-Code/blob/main/PRIVACY.md" ], "confidence": "medium", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 8, "scoreReason": "Roo Code’s archived repository is Apache-2.0 and remains available to inspect or fork. Its archive status means users must arrange ongoing fixes themselves. The code license does not determine the openness or terms of any connected model.", "argument": "Roo Code’s archived repository is Apache-2.0 and remains available to inspect or fork. Its archive status means users must arrange ongoing fixes themselves. The code license does not determine the openness or terms of any connected model.", "assessmentType": "evidence", "score": 8, "href": "/harnesses/roo-code/#openness" }, "labour": { "summary": "Roo Code’s archived repository exposes its former contribution trail, but public activity does not establish compensation or current maintainer support.", "sourceUrls": [ "https://github.com/RooCodeInc/Roo-Code" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "Roo Code’s archived repository preserves the former project’s contribution history, but does not show contributor compensation or current maintainer support. The archive makes past work inspectable while leaving labor conditions unknown. If you rely on a fork, budget for maintenance and ask its maintainers about their support model.", "argument": "Roo Code’s archived repository preserves the former project’s contribution history, but does not show contributor compensation or current maintainer support. The archive makes past work inspectable while leaving labor conditions unknown. If you rely on a fork, budget for maintenance and ask its maintainers about their support model.", "assessmentType": "conditional", "score": null, "href": "/harnesses/roo-code/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://roocodeinc.github.io/Roo-Code/providers/ollama/" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "Roo Code’s archived project materials do not identify a defense deployment. Its former provider choice means any government relationship would depend on which external model and host a user selected. Treat this as configuration-dependent, and verify the fork, provider, and host before use.", "argument": "Roo Code’s archived project materials do not identify a defense deployment. Its former provider choice means any government relationship would depend on which external model and host a user selected. Treat this as configuration-dependent, and verify the fork, provider, and host before use.", "assessmentType": "conditional", "score": null, "href": "/harnesses/roo-code/#military-contracts" }, "israel-palestine": { "summary": "We did not find Roo Code Inc. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Roo Code Inc. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/roo-code/#israel-palestine" } }, "sources": [ { "url": "https://github.com/RooCodeInc/Roo-Code", "title": "Roo Code archive or historical privacy notice", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/RooCodeInc/Roo-Code/blob/main/PRIVACY.md", "title": "Official privacy or data-processing disclosure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://roocodeinc.github.io/Roo-Code/providers/ollama/", "title": "Roo Code archived product local-model documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "releaseNote": "Historical profile: Roo Code announced its shutdown on May 15, 2025, and its source repository is archived. Privacy notes describe the former extension and may not represent any current service.", "href": "/harnesses/roo-code/", "overall": { "score": 8, "rated": 1, "total": 6, "coverage": 0.16666666666666666, "partial": true, "color": "green", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 2 } }, { "id": "aider", "name": "Aider", "category": "harnesses", "organization": "Aider-AI", "website": "https://aider.chat/", "summary": "Open-source terminal coding agent using hosted provider or local models.", "scope": "Local versus hosted execution and provider terms materially change privacy; Aider documents analytics and an anonymous identifier.", "dimensions": { "energy": { "summary": "Aider can route inference to a local model; that gives the user control over the serving machine, not a verified reduction in energy or water.", "sourceUrls": [ "https://aider.chat/docs/llms/ollama.html" ], "confidence": "medium", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Aider’s Ollama integration can keep inference on local hardware; its hosted-provider integrations leave energy accounting to those providers. The documentation includes no Aider task measurements, and local generation can consume more power when the model is large. Compare a small local model with a named cloud model on the same coding task.", "argument": "Aider’s Ollama integration can keep inference on local hardware; its hosted-provider integrations leave energy accounting to those providers. The documentation includes no Aider task measurements, and local generation can consume more power when the model is large. Compare a small local model with a named cloud model on the same coding task.", "assessmentType": "conditional", "score": null, "href": "/harnesses/aider/#energy" }, "training-rights": { "summary": "Aider is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://aider.chat/docs/llms", "https://aider.chat/docs/llms/ollama.html" ], "confidence": "medium", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "Aider’s local-model integration gives a user a way to inspect and run a chosen model, but Aider does not provide a training-data rights ledger for those models. A local endpoint avoids one provider’s data-handling terms; it does not prove the model’s source material was licensed. Select a model with a public dataset and license before sending work through it.", "argument": "Aider’s local-model integration gives a user a way to inspect and run a chosen model, but Aider does not provide a training-data rights ledger for those models. A local endpoint avoids one provider’s data-handling terms; it does not prove the model’s source material was licensed. Select a model with a public dataset and license before sending work through it.", "assessmentType": "conditional", "score": null, "href": "/harnesses/aider/#training-rights" }, "privacy": { "summary": "Aider can use local models or route code to a selected hosted provider; provider terms govern remote inference.", "sourceUrls": [ "https://aider.chat/docs/legal/privacy.html", "https://aider.chat/docs/llms", "https://aider.chat/docs/llms/ollama.html" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 6, "scoreReason": "The client exposes a practical local route and documents usage analytics, but it cannot override an API provider’s retention or training policy. Choose a local model for private work or check the selected provider’s exact data terms before sharing a repository.", "argument": "The client exposes a practical local route and documents usage analytics, but it cannot override an API provider’s retention or training policy. Choose a local model for private work or check the selected provider’s exact data terms before sharing a repository.", "assessmentType": "evidence", "score": 6, "href": "/harnesses/aider/#privacy" }, "openness": { "summary": "The Aider client is open source; selected inference model and host remain separate.", "sourceUrls": [ "https://github.com/Aider-AI/aider", "https://aider.chat/docs/legal/privacy.html" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 9, "scoreReason": "Aider’s public code is available under Apache-2.0, making the client inspectable and modifiable. The model and inference host remain separate choices with their own access terms. Open client code does not establish the training-data rights of a model.", "argument": "Aider’s public code is available under Apache-2.0, making the client inspectable and modifiable. The model and inference host remain separate choices with their own access terms. Open client code does not establish the training-data rights of a model.", "assessmentType": "evidence", "score": 9, "href": "/harnesses/aider/#openness" }, "labour": { "summary": "Aider’s public repository makes project contributions visible; it does not disclose contributor compensation or downstream model-labor conditions.", "sourceUrls": [ "https://github.com/Aider-AI/aider" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "Aider’s public repository shows a visible community contribution path, but not whether contributors receive compensation or support. That matters for sustainability as well as worker treatment. If adopting Aider in a business, consider funding its maintainers and ask your chosen model provider for annotation-labor disclosures.", "argument": "Aider’s public repository shows a visible community contribution path, but not whether contributors receive compensation or support. That matters for sustainability as well as worker treatment. If adopting Aider in a business, consider funding its maintainers and ask your chosen model provider for annotation-labor disclosures.", "assessmentType": "conditional", "score": null, "href": "/harnesses/aider/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://aider.chat/docs/llms/ollama.html" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "Aider is a model connector, and its docs list hosted providers as well as local models; the sources reviewed do not name an Aider defense contract. A provider’s military work is not evidence that Aider was used in it. If the relationship matters, select a local model or check each provider’s defense restrictions.", "argument": "Aider is a model connector, and its docs list hosted providers as well as local models; the sources reviewed do not name an Aider defense contract. A provider’s military work is not evidence that Aider was used in it. If the relationship matters, select a local model or check each provider’s defense restrictions.", "assessmentType": "conditional", "score": null, "href": "/harnesses/aider/#military-contracts" }, "israel-palestine": { "summary": "We did not find Aider-AI among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Aider-AI is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/aider/#israel-palestine" } }, "sources": [ { "url": "https://github.com/Aider-AI/aider", "title": "Official source repository or product page", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://aider.chat/docs/legal/privacy.html", "title": "Official privacy or data-processing disclosure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://aider.chat/docs/llms", "title": "Official product documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://aider.chat/docs/llms/ollama.html", "title": "Aider local Ollama configuration", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/harnesses/aider/", "overall": { "score": 7.5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 3 } }, { "id": "continue", "name": "Continue (read-only)", "category": "harnesses", "organization": "Continue", "website": "https://continue.dev/", "summary": "Open-source coding assistant and agent tooling for IDEs and CLI, configurable with hosted or local models.", "scope": "The final 2.0.0 release is available as a CLI, VS Code extension and JetBrains plugin, but the repository says development is read-only. Model-provider choice remains configurable; users who keep using it must own version checks and maintenance.", "dimensions": { "energy": { "summary": "Continue (archived) can route inference to a local model; that gives the user control over the serving machine, not a verified reduction in energy or water.", "sourceUrls": [ "https://docs.continue.dev/customize/model-providers/top-level/ollama" ], "confidence": "medium", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Continue’s current guide shows how to connect the editor to an Ollama model, including a remote Ollama endpoint. That makes the serving location a configuration choice, but the editor publishes no per-task energy or water measurements. Measure the selected model on its actual host before comparing it with a cloud route.", "argument": "Continue’s current guide shows how to connect the editor to an Ollama model, including a remote Ollama endpoint. That makes the serving location a configuration choice, but the editor publishes no per-task energy or water measurements. Measure the selected model on its actual host before comparing it with a cloud route.", "assessmentType": "conditional", "score": null, "href": "/harnesses/continue/#energy" }, "training-rights": { "summary": "Continue (archived) is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://docs.continue.dev/customize/models", "https://docs.continue.dev/customize/model-providers/top-level/ollama" ], "confidence": "medium", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "Continue documents both Ollama and a range of hosted model providers, but that provider guide does not disclose the training sources or permissions for each model. Running a model locally changes where inference happens; it does not establish rights in the model’s training data. Check the exact model’s provenance and license before choosing it.", "argument": "Continue documents both Ollama and a range of hosted model providers, but that provider guide does not disclose the training sources or permissions for each model. Running a model locally changes where inference happens; it does not establish rights in the model’s training data. Check the exact model’s provenance and license before choosing it.", "assessmentType": "conditional", "score": null, "href": "/harnesses/continue/#training-rights" }, "privacy": { "summary": "Continue supports local model configuration, while its project is read-only and some linked privacy guidance is version-dependent.", "sourceUrls": [ "https://docs.continue.dev/customize/deep-dives/development-data", "https://docs.continue.dev/guides/running-continue-without-internet" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 6, "scoreReason": "The final 2.0.0 release removed anonymous telemetry and authentication, while the offline guide describes older telemetry settings. Because the repository is read-only, versions and integrations may differ; verify the installed release and inspect the model route before sending code.", "argument": "The final 2.0.0 release removed anonymous telemetry and authentication, while the offline guide describes older telemetry settings. Because the repository is read-only, versions and integrations may differ; verify the installed release and inspect the model route before sending code.", "assessmentType": "evidence", "score": 6, "href": "/harnesses/continue/#privacy" }, "openness": { "summary": "Public source and Apache-2.0 license support inspection, but project maintenance is discontinued.", "sourceUrls": [ "https://github.com/continuedev/continue", "https://docs.continue.dev/customize/deep-dives/development-data" ], "confidence": "medium", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 8, "scoreReason": "Continue’s Apache-2.0 source can be inspected and modified, and the configuration supports different model providers. The project maintainers say the repository is read-only, so users who need ongoing fixes must maintain their own fork or choose an actively maintained alternative.", "argument": "Continue’s Apache-2.0 source can be inspected and modified, and the configuration supports different model providers. The project maintainers say the repository is read-only, so users who need ongoing fixes must maintain their own fork or choose an actively maintained alternative.", "assessmentType": "evidence", "score": 8, "href": "/harnesses/continue/#openness" }, "labour": { "summary": "Continue’s public repository shows community contribution history, but the project is read-only and does not disclose contributor compensation or model-data labor.", "sourceUrls": [ "https://github.com/continuedev/continue" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "The repository credits community contributors, but it does not report their compensation or working conditions. That public history documents participation, not whether contributors were paid or supported. Teams that depend on this read-only project should plan maintenance ownership and assess the selected model provider’s labor separately.", "argument": "The repository credits community contributors, but it does not report their compensation or working conditions. That public history documents participation, not whether contributors were paid or supported. Teams that depend on this read-only project should plan maintenance ownership and assess the selected model provider’s labor separately.", "assessmentType": "conditional", "score": null, "href": "/harnesses/continue/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://docs.continue.dev/customize/model-providers/top-level/ollama" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "Continue’s provider configuration allows local or hosted model routes, and the reviewed sources do not identify a Continue deployment in a defense workload. The model and inference host can have separate government relationships, which should not be attributed to the editor. Check the active provider and its specific contract scope.", "argument": "Continue’s provider configuration allows local or hosted model routes, and the reviewed sources do not identify a Continue deployment in a defense workload. The model and inference host can have separate government relationships, which should not be attributed to the editor. Check the active provider and its specific contract scope.", "assessmentType": "conditional", "score": null, "href": "/harnesses/continue/#military-contracts" }, "israel-palestine": { "summary": "We did not find Continue among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Continue is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/continue/#israel-palestine" } }, "sources": [ { "url": "https://github.com/continuedev/continue", "title": "Continue repository or documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.continue.dev/customize/deep-dives/development-data", "title": "Official license or documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.continue.dev/guides/running-continue-without-internet", "title": "Official product documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.continue.dev/customize/model-providers/top-level/ollama", "title": "Continue official guide to configuring a local Ollama provider", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.continue.dev/customize/models", "title": "Models", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "releaseNote": "Maintenance note: the official repository says it is no longer actively maintained and is read-only. The final 2.0.0 release removed anonymous telemetry and authentication; linked privacy guides may describe earlier versions, so confirm the version before relying on them.", "href": "/harnesses/continue/", "overall": { "score": 7, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 6 } }, { "id": "openhands", "name": "OpenHands Agent (Agent Canvas)", "category": "harnesses", "organization": "OpenHands", "website": "https://www.openhands.dev/", "summary": "OpenHands Agent is a model-configurable coding harness; Agent Canvas is the current local-first workspace for running it or other agents. The hosted Cloud service, local agent server, and selected model are separate data paths.", "scope": "This profile covers OpenHands Agent plus its current Agent Canvas client/workspace, not every agent that Canvas can connect to. Cloud privacy policy applies to hosted OpenHands services; self-hosted agent-server and local-model deployments are controlled by the operator.", "dimensions": { "energy": { "summary": "OpenHands can route inference to a local model; that gives the user control over the serving machine, not a verified reduction in energy or water.", "sourceUrls": [ "https://github.com/OpenHands/docs/blob/main/openhands/usage/llms/local-llms.mdx" ], "confidence": "medium", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "OpenHands documents local LLMs as well as hosted model profiles, so an operator can choose where inference runs. It publishes no energy per task, and its local-model guide notes substantial hardware requirements for capable agentic models. Measure the server and full agent run, including retries, before preferring local deployment on energy grounds.", "argument": "OpenHands documents local LLMs as well as hosted model profiles, so an operator can choose where inference runs. It publishes no energy per task, and its local-model guide notes substantial hardware requirements for capable agentic models. Measure the server and full agent run, including retries, before preferring local deployment on energy grounds.", "assessmentType": "conditional", "score": null, "href": "/harnesses/openhands/#energy" }, "training-rights": { "summary": "OpenHands’ hosted-service privacy policy says user content and feedback may be used to train or tune its AI models; local-model support gives a separate route.", "sourceUrls": [ "https://www.openhands.dev/privacy", "https://github.com/OpenHands/docs/blob/main/openhands/usage/llms/local-llms.mdx" ], "confidence": "high", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": 3, "scoreReason": "OpenHands Cloud’s privacy policy permits use of content and feedback to train or tune service models, while local LLMs are documented separately. Running a local model avoids that particular hosted-content path but does not establish that model’s data was used with permission. For cloud use, ask for a written training exclusion; for local use, inspect the model’s dataset and license.", "argument": "OpenHands Cloud’s privacy policy permits use of content and feedback to train or tune service models, while local LLMs are documented separately. Running a local model avoids that particular hosted-content path but does not establish that model’s data was used with permission. For cloud use, ask for a written training exclusion; for local use, inspect the model’s dataset and license.", "assessmentType": "evidence", "score": 3, "href": "/harnesses/openhands/#training-rights" }, "privacy": { "summary": "OpenHands’ current hosted-service policy allows content and feedback to be used to train or tune its AI models and retains content as needed for business/legal purposes.", "sourceUrls": [ "https://www.openhands.dev/", "https://www.openhands.dev/privacy" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 3, "scoreReason": "The cloud policy allows model-training use and does not show a training opt-out in the reviewed text; this is a material privacy limit. OpenHands documents local/self-hosted deployment and local LLMs, which shifts control to the operator but also makes the operator responsible for securing logs, tools, and the model endpoint.", "argument": "The cloud policy allows model-training use and does not show a training opt-out in the reviewed text; this is a material privacy limit. OpenHands documents local/self-hosted deployment and local LLMs, which shifts control to the operator but also makes the operator responsible for securing logs, tools, and the model endpoint.", "assessmentType": "evidence", "score": 3, "href": "/harnesses/openhands/#privacy" }, "openness": { "summary": "Public repository supports inspection; this does not establish openness of every hosted or enterprise component.", "sourceUrls": [ "https://github.com/OpenHands/OpenHands", "https://www.openhands.dev/", "https://www.openhands.dev/product/canvas", "https://github.com/OpenHands/OpenHands/releases" ], "confidence": "medium", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 8, "scoreReason": "OpenHands publishes its agent code under a permissive license, which allows inspection and modification of the client. Hosted and enterprise services may include components outside that repository. Check the service and model terms separately from the open-source code.", "argument": "OpenHands publishes its agent code under a permissive license, which allows inspection and modification of the client. Hosted and enterprise services may include components outside that repository. Check the service and model terms separately from the open-source code.", "assessmentType": "evidence", "score": 8, "href": "/harnesses/openhands/#openness" }, "labour": { "summary": "OpenHands has a public contributor repository, while its privacy policy names service providers; neither source provides worker pay or conditions for model data work.", "sourceUrls": [ "https://github.com/OpenHands/OpenHands", "https://www.openhands.dev/privacy" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "OpenHands describes Agent Canvas as community-built and publishes its source, but neither that statement nor the privacy policy reports contributor pay or model-data labor conditions. Community involvement provides participation, not worker-level assurance. Ask the project and selected model provider for current compensation and contractor safeguards.", "argument": "OpenHands describes Agent Canvas as community-built and publishes its source, but neither that statement nor the privacy policy reports contributor pay or model-data labor conditions. Community involvement provides participation, not worker-level assurance. Ask the project and selected model provider for current compensation and contractor safeguards.", "assessmentType": "conditional", "score": null, "href": "/harnesses/openhands/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://github.com/OpenHands/docs/blob/main/openhands/usage/llms/local-llms.mdx" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "OpenHands Agent Canvas can use a local backend, OpenHands Cloud, or another agent, and the sources reviewed do not name a defense deployment. That flexibility means the selected model and host determine much of the exposure. Check the backend and provider contract before deploying it in a defense-sensitive environment.", "argument": "OpenHands Agent Canvas can use a local backend, OpenHands Cloud, or another agent, and the sources reviewed do not name a defense deployment. That flexibility means the selected model and host determine much of the exposure. Check the backend and provider contract before deploying it in a defense-sensitive environment.", "assessmentType": "conditional", "score": null, "href": "/harnesses/openhands/#military-contracts" }, "israel-palestine": { "summary": "We did not find OpenHands among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "OpenHands is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/openhands/#israel-palestine" } }, "sources": [ { "url": "https://github.com/OpenHands/OpenHands", "title": "Official source repository or product page", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.openhands.dev/", "title": "Official license or documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/OpenHands/docs/blob/main/openhands/usage/llms/local-llms.mdx", "title": "OpenHands local LLM configuration", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.openhands.dev/privacy", "title": "OpenHands Privacy Policy (effective 2025-09-03)", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.openhands.dev/product/canvas", "title": "OpenHands Agent Canvas product overview", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/OpenHands/OpenHands/releases", "title": "OpenHands repository releases; v1.30.1 dated 2026-10-22", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "releaseNote": "Current product naming: OpenHands describes Agent Canvas as its new local workspace and OpenHands Agent as one supported harness. The current repository release list shows v1.30.1 dated 2026-10-22.", "href": "/harnesses/openhands/", "overall": { "score": 4.7, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 11 } }, { "id": "goose", "name": "Goose", "category": "harnesses", "organization": "Block / AAIF", "website": "https://block.github.io/goose/", "summary": "Open-source local-first desktop and CLI coding agent supporting hosted and local providers.", "scope": "Local installation does not mean inference is local unless a local provider is selected.", "dimensions": { "energy": { "summary": "Goose can route inference to a local model; that gives the user control over the serving machine, not a verified reduction in energy or water.", "sourceUrls": [ "https://block.github.io/goose/" ], "confidence": "medium", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Goose runs its agent on the user’s machine and supports both Ollama and hosted model providers. That makes a local route possible, but does not tell us whether a full agent session uses less electricity than a hosted model. Choose a task-appropriate small model, record local energy, and compare with provider-specific emissions data.", "argument": "Goose runs its agent on the user’s machine and supports both Ollama and hosted model providers. That makes a local route possible, but does not tell us whether a full agent session uses less electricity than a hosted model. Choose a task-appropriate small model, record local energy, and compare with provider-specific emissions data.", "assessmentType": "conditional", "score": null, "href": "/harnesses/goose/#energy" }, "training-rights": { "summary": "Goose is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://block.github.io/goose/" ], "confidence": "medium", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "Goose offers hosted providers and local Ollama, but its own documentation does not publish training rights for every model. Provider choice matters: a local model can have a public corpus and license, while a hosted API follows its provider’s terms. Check the selected model’s source and rights records rather than treating the open-source agent as an open-data guarantee.", "argument": "Goose offers hosted providers and local Ollama, but its own documentation does not publish training rights for every model. Provider choice matters: a local model can have a public corpus and license, while a hosted API follows its provider’s terms. Check the selected model’s source and rights records rather than treating the open-source agent as an open-data guarantee.", "assessmentType": "conditional", "score": null, "href": "/harnesses/goose/#training-rights" }, "privacy": { "summary": "Goose runs locally as a harness, but prompts still go to the selected model provider unless a local model is configured.", "sourceUrls": [ "https://block.github.io/goose/" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 6, "scoreReason": "Its provider list includes Ollama as well as hosted vendors, so “local agent” should not be read as “local inference.” Select a local model or review the chosen provider’s retention and training terms, and audit any enabled extension separately.", "argument": "Goose runs its agent locally, but that does not mean model inference stays on the device. Its provider list includes Ollama as well as hosted vendors, so the selected route controls where prompts go and which policy applies. Choose a local model for sensitive work or review the provider’s retention terms, and audit each extension separately.", "assessmentType": "evidence", "score": 6, "href": "/harnesses/goose/#privacy" }, "openness": { "summary": "Goose repository is Apache-2.0; provider choice is separate from model openness.", "sourceUrls": [ "https://github.com/aaif-goose/goose", "https://block.github.io/goose/" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 9, "scoreReason": "Goose’s Apache-2.0 source can be inspected and modified, while its provider integrations connect to separately licensed models and services. This gives users control over the client code without opening every part of the inference stack. Review the selected model’s availability and terms.", "argument": "Goose’s Apache-2.0 source can be inspected and modified, while its provider integrations connect to separately licensed models and services. This gives users control over the client code without opening every part of the inference stack. Review the selected model’s availability and terms.", "assessmentType": "evidence", "score": 9, "href": "/harnesses/goose/#openness" }, "labour": { "summary": "Goose advertises a large open-source contributor community and AAIF affiliation; neither fact supplies worker-level pay or contractor evidence.", "sourceUrls": [ "https://github.com/aaif-goose/goose", "https://block.github.io/goose/" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "Goose reports hundreds of contributors and AAIF affiliation, which indicates broad participation and shared stewardship. It does not establish contributor pay or the working conditions at the model providers Goose connects to. Support project maintainers where appropriate and ask the selected provider for worker-level evidence.", "argument": "Goose reports hundreds of contributors and AAIF affiliation, which indicates broad participation and shared stewardship. It does not establish contributor pay or the working conditions at the model providers Goose connects to. Support project maintainers where appropriate and ask the selected provider for worker-level evidence.", "assessmentType": "conditional", "score": null, "href": "/harnesses/goose/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://block.github.io/goose/" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "Goose’s provider choices include local models and commercial APIs, and no reviewed source ties Goose itself to a defense contract. A model company’s government work does not mean Goose was used. Choose a local model when the provider relationship is unacceptable, and check extension vendors separately.", "argument": "Goose’s provider choices include local models and commercial APIs, and no reviewed source ties Goose itself to a defense contract. A model company’s government work does not mean Goose was used. Choose a local model when the provider relationship is unacceptable, and check extension vendors separately.", "assessmentType": "conditional", "score": null, "href": "/harnesses/goose/#military-contracts" }, "israel-palestine": { "summary": "We did not find Block / AAIF among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Block / AAIF is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/goose/#israel-palestine" } }, "sources": [ { "url": "https://github.com/aaif-goose/goose", "title": "Official source repository or product page", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://block.github.io/goose/", "title": "Goose provider and local model overview", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/harnesses/goose/", "overall": { "score": 7.5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 4 } }, { "id": "opencode", "name": "OpenCode", "category": "harnesses", "organization": "Anomaly", "website": "https://opencode.ai/", "summary": "Open-source terminal/desktop coding agent with configurable providers and agent permissions.", "scope": "Open harness with hosted or local inference; security guidance says local execution is not itself a sandbox.", "dimensions": { "energy": { "summary": "OpenCode can route inference to a local model; that gives the user control over the serving machine, not a verified reduction in energy or water.", "sourceUrls": [ "https://opencode.ai/v2/docs/models" ], "confidence": "medium", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "OpenCode discovers local Ollama, LM Studio, and vLLM models as well as hosted providers. Local inference lets the operator measure power draw, but long context and repeated tool calls can make the whole task energy-intensive. Benchmark a completed coding task at the same quality level and include retries and context size.", "argument": "OpenCode discovers local Ollama, LM Studio, and vLLM models as well as hosted providers. Local inference lets the operator measure power draw, but long context and repeated tool calls can make the whole task energy-intensive. Benchmark a completed coding task at the same quality level and include retries and context size.", "assessmentType": "conditional", "score": null, "href": "/harnesses/opencode/#energy" }, "training-rights": { "summary": "OpenCode is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://opencode.ai/v2/docs/models" ], "confidence": "medium", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "OpenCode’s local model discovery lets a user choose a model they can inspect, while hosted providers remain governed by their own data terms. The client does not publish a training corpus or rights ledger for the models it lists. Choose a model with source-level provenance and a clear license, and review the host if inference is remote.", "argument": "OpenCode’s local model discovery lets a user choose a model they can inspect, while hosted providers remain governed by their own data terms. The client does not publish a training corpus or rights ledger for the models it lists. Choose a model with source-level provenance and a clear license, and review the host if inference is remote.", "assessmentType": "conditional", "score": null, "href": "/harnesses/opencode/#training-rights" }, "privacy": { "summary": "OpenCode supports local models and many hosted providers, with data handling determined by the chosen route.", "sourceUrls": [ "https://github.com/anomalyco/opencode/security", "https://opencode.ai/v2/docs/models" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 6, "scoreReason": "The model selector and local-server support give users real routing choice, but an open client does not protect prompts sent to a remote provider. Use a local endpoint for sensitive repositories or review provider retention before use.", "argument": "The model selector and local-server support give users real routing choice, but an open client does not protect prompts sent to a remote provider. Use a local endpoint for sensitive repositories or review provider retention before use.", "assessmentType": "evidence", "score": 6, "href": "/harnesses/opencode/#privacy" }, "openness": { "summary": "MIT-licensed source is inspectable and modifiable; this does not open provider models.", "sourceUrls": [ "https://github.com/anomalyco/opencode", "https://github.com/anomalyco/opencode/security" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 9, "scoreReason": "OpenCode’s MIT-licensed repository makes the client inspectable and modifiable. Its supported model providers remain separate services with their own access conditions. An open client does not grant access to model weights or training records.", "argument": "OpenCode’s MIT-licensed repository makes the client inspectable and modifiable. Its supported model providers remain separate services with their own access conditions. An open client does not grant access to model weights or training records.", "assessmentType": "evidence", "score": 9, "href": "/harnesses/opencode/#openness" }, "labour": { "summary": "OpenCode’s public source exposes contributions but does not document worker compensation, annotation labor, or provider work conditions.", "sourceUrls": [ "https://github.com/anomalyco/opencode" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "OpenCode’s public repository lets users inspect contributions, but the sources reviewed do not describe how maintainers are funded or how model-data workers are treated. Open-source access improves visibility into software development, not the full labor chain. Ask the project about maintainer support and the model provider about data-worker protections.", "argument": "OpenCode’s public repository lets users inspect contributions, but the sources reviewed do not describe how maintainers are funded or how model-data workers are treated. Open-source access improves visibility into software development, not the full labor chain. Ask the project about maintainer support and the model provider about data-worker protections.", "assessmentType": "conditional", "score": null, "href": "/harnesses/opencode/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://opencode.ai/v2/docs/models" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "OpenCode can use local servers or hosted providers, and its reviewed public materials do not identify a defense deployment. The relevant contract, if any, belongs to the selected model or infrastructure provider unless OpenCode use is documented. Verify the selected stack and its restrictions before using it for sensitive work.", "argument": "OpenCode can use local servers or hosted providers, and its reviewed public materials do not identify a defense deployment. The relevant contract, if any, belongs to the selected model or infrastructure provider unless OpenCode use is documented. Verify the selected stack and its restrictions before using it for sensitive work.", "assessmentType": "conditional", "score": null, "href": "/harnesses/opencode/#military-contracts" }, "israel-palestine": { "summary": "We did not find Anomaly among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Anomaly is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/opencode/#israel-palestine" } }, "sources": [ { "url": "https://github.com/anomalyco/opencode", "title": "Official source repository or product page", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/anomalyco/opencode/security", "title": "Official security guidance", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://opencode.ai/v2/docs/models", "title": "OpenCode hosted and local model configuration", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/harnesses/opencode/", "overall": { "score": 7.5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 5 } }, { "id": "kilo-code", "name": "Kilo Code", "category": "harnesses", "organization": "Kilo", "website": "https://kilo.ai/", "summary": "Open-source coding harness with IDE, CLI, and cloud product paths and broad model-provider support.", "scope": "Current Kilo platform and legacy IDE are distinct; legacy IDE reached end of life.", "dimensions": { "energy": { "summary": "Kilo Code can route inference to a local model; that gives the user control over the serving machine, not a verified reduction in energy or water.", "sourceUrls": [ "https://kilo.ai/docs/ai-providers/ollama" ], "confidence": "medium", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Kilo supports local Ollama and hosted providers; its guide notes that larger contexts use more memory and can affect performance. The product publishes no emissions comparison, and local routing alone says nothing about the electricity source or workload footprint. Measure the local device use against a named hosted alternative.", "argument": "Kilo supports local Ollama and hosted providers; its guide notes that larger contexts use more memory and can affect performance. The product publishes no emissions comparison, and local routing alone says nothing about the electricity source or workload footprint. Measure the local device use against a named hosted alternative.", "assessmentType": "conditional", "score": null, "href": "/harnesses/kilo-code/#energy" }, "training-rights": { "summary": "Kilo Code is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://kilo.ai/privacy", "https://kilo.ai/docs/ai-providers/ollama" ], "confidence": "medium", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "Kilo supports local models and hosted gateways, but these routes answer different questions: where prompts go and whether the model’s training inputs were permitted. Local inference may improve control over prompts without resolving corpus rights. For rights-sensitive projects, use a model with documented training sources or get the provider’s terms in writing.", "argument": "Kilo supports local models and hosted gateways, but these routes answer different questions: where prompts go and whether the model’s training inputs were permitted. Local inference may improve control over prompts without resolving corpus rights. For rights-sensitive projects, use a model with documented training sources or get the provider’s terms in writing.", "assessmentType": "conditional", "score": null, "href": "/harnesses/kilo-code/#training-rights" }, "privacy": { "summary": "Kilo’s privacy policy says hosted requests are routed to the selected provider and downstream retention may differ; local Ollama is also supported.", "sourceUrls": [ "https://kilo.ai/privacy", "https://github.com/Kilo-Org/kilocode/blob/main/PRIVACY.md" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 5, "scoreReason": "Deleting Kilo-side history does not force the model provider to delete its copy, so hosted routing remains a meaningful exception. For private work, configure a local provider and disable any unrelated cloud or gateway features you do not need.", "argument": "Deleting Kilo-side history does not force the model provider to delete its copy, so hosted routing remains a meaningful exception. For private work, configure a local provider and disable any unrelated cloud or gateway features you do not need.", "assessmentType": "evidence", "score": 5, "href": "/harnesses/kilo-code/#privacy" }, "openness": { "summary": "Current repository is MIT-licensed; hosted services and providers remain separate.", "sourceUrls": [ "https://github.com/Kilo-Org/kilocode", "https://github.com/Kilo-Org/kilocode-legacy" ], "confidence": "medium", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 8, "scoreReason": "Kilo Code’s current repository is MIT-licensed and can be inspected or modified. Its hosted services and connected models remain separate from that codebase. The legacy IDE extension has a distinct repository and lifecycle.", "argument": "Kilo Code’s current repository is MIT-licensed and can be inspected or modified. Its hosted services and connected models remain separate from that codebase. The legacy IDE extension has a distinct repository and lifecycle.", "assessmentType": "evidence", "score": 8, "href": "/harnesses/kilo-code/#openness" }, "labour": { "summary": "Kilo’s public code and privacy policy describe product/provider architecture, not the labor conditions of maintainers, vendors, or model-data workers.", "sourceUrls": [ "https://github.com/Kilo-Org/kilocode", "https://kilo.ai/privacy" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "Kilo’s public code and privacy documentation explain product and provider paths, but not the pay or conditions of maintainers, vendors, or data workers. The product spans local clients and hosted services, so a single source would not cover the whole labor chain. Ask Kilo and the selected model provider for current worker and vendor disclosures.", "argument": "Kilo’s public code and privacy documentation explain product and provider paths, but not the pay or conditions of maintainers, vendors, or data workers. The product spans local clients and hosted services, so a single source would not cover the whole labor chain. Ask Kilo and the selected model provider for current worker and vendor disclosures.", "assessmentType": "conditional", "score": null, "href": "/harnesses/kilo-code/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://kilo.ai/docs/ai-providers/ollama" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "Kilo supports both local inference and hosted model gateways; the sources reviewed do not identify Kilo Code in a defense deployment. Gateway or model-provider government work is a separate relationship unless the specific Kilo service is named. Choose a local route or ask Kilo which providers serve the selected model.", "argument": "Kilo supports both local inference and hosted model gateways; the sources reviewed do not identify Kilo Code in a defense deployment. Gateway or model-provider government work is a separate relationship unless the specific Kilo service is named. Choose a local route or ask Kilo which providers serve the selected model.", "assessmentType": "conditional", "score": null, "href": "/harnesses/kilo-code/#military-contracts" }, "israel-palestine": { "summary": "We did not find Kilo among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Kilo is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/kilo-code/#israel-palestine" } }, "sources": [ { "url": "https://github.com/Kilo-Org/kilocode", "title": "Official source repository or product page", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/Kilo-Org/kilocode-legacy", "title": "Official license or documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://kilo.ai/privacy", "title": "Kilo privacy policy and provider routing", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/Kilo-Org/kilocode/blob/main/PRIVACY.md", "title": "Official privacy or data-processing disclosure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://kilo.ai/docs/ai-providers/ollama", "title": "Kilo Code local Ollama setup", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "releaseNote": "Product distinction: this entry covers the current Kilo Code repository and platform. Kilo Code Legacy is a separate IDE extension whose repository records end of life on July 31, 2026.", "href": "/harnesses/kilo-code/", "overall": { "score": 6.5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 7 } }, { "id": "devin", "name": "Devin", "category": "harnesses", "organization": "Cognition", "website": "https://devin.ai/", "summary": "Hosted autonomous software-engineering agent operating in a managed development environment.", "scope": "Cognition DPA describes processors and processing; all retention settings, model paths, and labor outcomes are not established here.", "dimensions": { "energy": { "summary": "Devin does not publish service-specific energy or water data; the model and serving provider determine most inference impact.", "sourceUrls": [ "https://app.devin.ai/" ], "confidence": "low", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Devin runs sessions in Cognition-managed virtual machines and offers model choices, but the current product materials disclose no serving-region or per-task energy figures. Session isolation is a security control, not an energy measurement. Ask Cognition for model, region, VM, and workload data, then compare a completed task rather than a token-price proxy.", "argument": "Devin runs sessions in Cognition-managed virtual machines and offers model choices, but the current product materials disclose no serving-region or per-task energy figures. Session isolation is a security control, not an energy measurement. Ask Cognition for model, region, VM, and workload data, then compare a completed task rather than a token-price proxy.", "assessmentType": "argument", "score": null, "href": "/harnesses/devin/#energy" }, "training-rights": { "summary": "Devin is a harness that connects to models; its sources do not establish a complete rights record for the training data of every model it can use.", "sourceUrls": [ "https://app.devin.ai/", "https://cognition.ai/Cognition-DPA.pdf" ], "confidence": "low", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": null, "scoreReason": "Devin’s managed agent uses selected cloud models and Cognition says it does not train on customer business data; this does not disclose the underlying models’ pretraining data. The cited product sources do not document local model execution or a corpus-level rights audit. Ask Cognition which model version serves your session and review its training provenance before relying on it.", "argument": "Devin’s managed agent uses selected cloud models and Cognition says it does not train on customer business data; this does not disclose the underlying models’ pretraining data. The cited product sources do not document local model execution or a corpus-level rights audit. Ask Cognition which model version serves your session and review its training provenance before relying on it.", "assessmentType": "conditional", "score": null, "href": "/harnesses/devin/#training-rights" }, "privacy": { "summary": "Devin’s enterprise page says provider zero-retention is enforced and customer data is not used for model training, with customer retention controls.", "sourceUrls": [ "https://cognition.ai/Cognition-DPA.pdf", "https://app.devin.ai/" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 7, "scoreReason": "These are strong hosted-service commitments, while each session still runs in Cognition’s managed VM unless the plan provides a customer-VPC option. Confirm the contract, retention window, and exact provider/model route for your plan before uploading sensitive code.", "argument": "These are strong hosted-service commitments, while each session still runs in Cognition’s managed VM unless the plan provides a customer-VPC option. Confirm the contract, retention window, and exact provider/model route for your plan before uploading sensitive code.", "assessmentType": "evidence", "score": 7, "href": "/harnesses/devin/#privacy" }, "openness": { "summary": "Hosted proprietary service; public source or agent weights are not established.", "sourceUrls": [ "https://docs.devin.ai/", "https://cognition.ai/Cognition-DPA.pdf" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 2, "scoreReason": "Devin is a hosted commercial agent service; the reviewed product and data-processing documents do not publish its implementation under an open-source license. Users can evaluate the service through its controls and terms, but cannot audit the complete agent stack from public source. Model and hosting details remain separate questions.", "argument": "Devin is a hosted commercial agent service; the reviewed product and data-processing documents do not publish its implementation under an open-source license. Users can evaluate the service through its controls and terms, but cannot audit the complete agent stack from public source. Model and hosting details remain separate questions.", "assessmentType": "evidence", "score": 2, "href": "/harnesses/devin/#openness" }, "labour": { "summary": "Cognition publishes processing and enterprise security material, but the sources reviewed do not disclose worker conditions in model training or service support.", "sourceUrls": [ "https://cognition.ai/Cognition-DPA.pdf", "https://app.devin.ai/" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "Cognition’s DPA identifies service processors and Devin advertises enterprise controls, but neither source reports the labor conditions of model-data workers or support contractors. Security and privacy controls do not answer worker-treatment questions. Request Cognition’s supplier standards, worker locations, and independent review for the specific service chain.", "argument": "Cognition’s DPA identifies service processors and Devin advertises enterprise controls, but neither source reports the labor conditions of model-data workers or support contractors. Security and privacy controls do not answer worker-treatment questions. Request Cognition’s supplier standards, worker locations, and independent review for the specific service chain.", "assessmentType": "argument", "score": null, "href": "/harnesses/devin/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://app.devin.ai/" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "Devin is a hosted managed agent, but the sources reviewed do not identify a military customer or defense workload. Because there is no local model path established here, the service’s cloud provider and Cognition’s own contracts are the relevant checks. Ask Cognition directly about defense use and obtain the applicable customer terms.", "argument": "Devin is a hosted managed agent, but the sources reviewed do not identify a military customer or defense workload. Because there is no local model path established here, the service’s cloud provider and Cognition’s own contracts are the relevant checks. Ask Cognition directly about defense use and obtain the applicable customer terms.", "assessmentType": "conditional", "score": null, "href": "/harnesses/devin/#military-contracts" }, "israel-palestine": { "summary": "We did not find Cognition among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Cognition is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/devin/#israel-palestine" } }, "sources": [ { "url": "https://docs.devin.ai/", "title": "Official source repository or product page", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cognition.ai/Cognition-DPA.pdf", "title": "Cognition Data Processing Agreement and subprocessors", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://app.devin.ai/", "title": "Devin agent service and enterprise controls", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/harnesses/devin/", "overall": { "score": 4.5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 12 } }, { "id": "replit-agent", "name": "Replit Agent", "category": "harnesses", "organization": "Replit", "website": "https://replit.com/ai", "summary": "Hosted app-building and coding agent integrated with Replit’s browser development platform.", "scope": "Prompts, code, and projects are processed by hosted Replit service; current policy and older engineering post differ in scope.", "dimensions": { "energy": { "summary": "Replit Agent does not publish service-specific energy or water data; the model and serving provider determine most inference impact.", "sourceUrls": [ "https://docs.replit.com/replitai/assistant/" ], "confidence": "low", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Replit Agent is a hosted service with Lite, Economy, and Power modes; the docs describe cost and speed tradeoffs, not electricity or water. A cheaper credit mode is not automatically a lower-emission mode. Use the lowest mode that completes the task and request Replit’s model- and region-specific energy data before comparing it.", "argument": "Replit Agent is a hosted service with Lite, Economy, and Power modes; the docs describe cost and speed tradeoffs, not electricity or water. A cheaper credit mode is not automatically a lower-emission mode. Use the lowest mode that completes the task and request Replit’s model- and region-specific energy data before comparing it.", "assessmentType": "argument", "score": null, "href": "/harnesses/replit-agent/#energy" }, "training-rights": { "summary": "Replit’s current privacy policy allows machine-learning improvement uses; an older engineering post says non-public code was excluded from training, leaving the exact scope unclear.", "sourceUrls": [ "https://replit.com/privacy-policy", "https://frontier.replit.com/how-replit-makes-sense-of-code-at-scale-ai-data/" ], "confidence": "medium", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": 4, "scoreReason": "Replit’s hosted Agent offers task modes, but the cited docs do not provide a local inference route or a corpus-level rights record for its models. Its privacy policy allows machine-learning improvement uses, and the older private-code assurance has a different scope. Ask which Agent data may be used for training on your plan and request a written opt-out.", "argument": "Replit’s hosted Agent offers task modes, but the cited docs do not provide a local inference route or a corpus-level rights record for its models. Its privacy policy allows machine-learning improvement uses, and the older private-code assurance has a different scope. Ask which Agent data may be used for training on your plan and request a written opt-out.", "assessmentType": "evidence", "score": 4, "href": "/harnesses/replit-agent/#training-rights" }, "privacy": { "summary": "Replit’s current policy collects prompts, code, and project content and permits machine-learning improvement uses; an older engineering post describes a narrower training boundary.", "sourceUrls": [ "https://replit.com/privacy-policy", "https://frontier.replit.com/how-replit-makes-sense-of-code-at-scale-ai-data/" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 4, "scoreReason": "The current policy is broader than the earlier assurance that private code was excluded from training, and the sources do not reconcile those statements. Because the Agent is hosted, ask Replit which content is used for model improvement on your plan and how to opt out before working with private repositories.", "argument": "The current policy is broader than the earlier assurance that private code was excluded from training, and the sources do not reconcile those statements. Because the Agent is hosted, ask Replit which content is used for model improvement on your plan and how to opt out before working with private repositories.", "assessmentType": "evidence", "score": 4, "href": "/harnesses/replit-agent/#privacy" }, "openness": { "summary": "Hosted proprietary product; source and agent implementation are not publicly inspectable.", "sourceUrls": [ "https://replit.com/privacy-policy", "https://frontier.replit.com/how-replit-makes-sense-of-code-at-scale-ai-data/" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 2, "scoreReason": "Replit Agent is offered as a hosted commercial product, and the reviewed product and privacy materials do not publish its implementation under an open-source license. Users can inspect product terms and data controls, but not the full agent implementation. Model and hosting components may have separate terms.", "argument": "Replit Agent is offered as a hosted commercial product, and the reviewed product and privacy materials do not publish its implementation under an open-source license. Users can inspect product terms and data controls, but not the full agent implementation. Model and hosting components may have separate terms.", "assessmentType": "evidence", "score": 2, "href": "/harnesses/replit-agent/#openness" }, "labour": { "summary": "Replit explains service data use, but the material reviewed does not name contractors, pay, protections, or grievance routes for model data work.", "sourceUrls": [ "https://replit.com/privacy-policy", "https://replit.com/" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "Replit’s privacy policy explains data use but not who annotates or evaluates its models, what those workers are paid, or what protections they have. This leaves the labor impact of the hosted Agent hard to assess. Ask Replit for the current data-work vendors, safety provisions, and independent labor audit results.", "argument": "Replit’s privacy policy explains data use but not who annotates or evaluates its models, what those workers are paid, or what protections they have. This leaves the labor impact of the hosted Agent hard to assess. Ask Replit for the current data-work vendors, safety provisions, and independent labor audit results.", "assessmentType": "argument", "score": null, "href": "/harnesses/replit-agent/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://docs.replit.com/replitai/assistant/" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "Replit Agent is a hosted coding service, but the sources reviewed do not name it in a defense or surveillance deployment. A model-provider contract alone would not show that Replit Agent was part of it. Ask Replit about government customers and contract restrictions if that relationship affects your choice.", "argument": "Replit Agent is a hosted coding service, but the sources reviewed do not name it in a defense or surveillance deployment. A model-provider contract alone would not show that Replit Agent was part of it. Ask Replit about government customers and contract restrictions if that relationship affects your choice.", "assessmentType": "conditional", "score": null, "href": "/harnesses/replit-agent/#military-contracts" }, "israel-palestine": { "summary": "We did not find Replit among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Replit is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/replit-agent/#israel-palestine" } }, "sources": [ { "url": "https://replit.com/privacy-policy", "title": "Replit Privacy Policy (updated 2026-10-21)", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://frontier.replit.com/how-replit-makes-sense-of-code-at-scale-ai-data/", "title": "Replit engineering post on code analytics and training", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.replit.com/replitai/assistant/", "title": "Replit Agent modes and cost controls", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://replit.com/", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/harnesses/replit-agent/", "overall": { "score": 3.3, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "red", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 17 } }, { "id": "zed", "name": "Zed", "category": "harnesses", "organization": "Zed Industries", "website": "https://zed.dev/", "summary": "Open-source editor with integrated agent features, hosted models, BYO provider, and external-agent support.", "scope": "Privacy differs by request path; hosted provider terms have designated safety-retention exceptions, while BYO/external agents follow their terms.", "dimensions": { "energy": { "summary": "Zed can route inference to a local model; that gives the user control over the serving machine, not a verified reduction in energy or water.", "sourceUrls": [ "https://zed.dev/docs/ai/use-a-local-model" ], "confidence": "medium", "unknowns": "No per-task electricity, water, or emissions data, serving-region breakdown, or lifecycle assessment was found for the harness.", "suggestedScore": null, "scoreReason": "Zed can send inference to hosted models or local Ollama/LM Studio servers. Local use makes device electricity measurable, but Zed publishes no comparison between that path and hosted inference. Choose a compact model where suitable and measure a complete editing task—including context and retries—before claiming an energy benefit.", "argument": "Zed can send inference to hosted models or local Ollama/LM Studio servers. Local use makes device electricity measurable, but Zed publishes no comparison between that path and hosted inference. Choose a compact model where suitable and measure a complete editing task—including context and retries—before claiming an energy benefit.", "assessmentType": "conditional", "score": null, "href": "/harnesses/zed/#energy" }, "training-rights": { "summary": "Zed documents opt-in Edit Prediction collection from open-source projects and publishes information about its Zeta model dataset; BYO models remain outside this policy.", "sourceUrls": [ "https://zed.dev/docs/ai/ai-improvement", "https://zed.dev/blog/zeta2" ], "confidence": "high", "unknowns": "The provider and exact model vary by configuration; no complete rights ledger can be inferred from the harness license or provider list.", "suggestedScore": 8, "scoreReason": "Zed’s own Edit Prediction training collection is opt-in and restricted to open-source projects, and its Zeta2 page describes the dataset approach. That is meaningful evidence for Zed’s own model, not a blanket endorsement of BYO providers or external agents. Prefer the Zed model if those controls fit, and audit each outside model separately.", "argument": "Zed’s own Edit Prediction training collection is opt-in and restricted to open-source projects, and its Zeta2 page describes the dataset approach. That is meaningful evidence for Zed’s own model, not a blanket endorsement of BYO providers or external agents. Prefer the Zed model if those controls fit, and audit each outside model separately.", "assessmentType": "evidence", "score": 8, "href": "/harnesses/zed/#training-rights" }, "privacy": { "summary": "Zed says ordinary requests are not stored or trained on, with provider safety-retention exceptions and explicit opt-ins for feedback/training.", "sourceUrls": [ "https://zed.dev/docs/ai/privacy-and-security", "https://zed.dev/docs/business/privacy" ], "confidence": "high", "unknowns": "Plan, selected model/provider, enabled features, and deployment mode can change this assessment.", "suggestedScore": 8, "scoreReason": "The product gives clear controls for its hosted model path, though Anthropic-designated models may retain prompts for safety and BYO/external agents follow their own terms. Choose an eligible no-retention model or local provider if provider-side retention is unacceptable.", "argument": "The product gives clear controls for its hosted model path, though Anthropic-designated models may retain prompts for safety and BYO/external agents follow their own terms. Choose an eligible no-retention model or local provider if provider-side retention is unacceptable.", "assessmentType": "evidence", "score": 8, "href": "/harnesses/zed/#privacy" }, "openness": { "summary": "Editor is open source under GPL and server components AGPL; hosted AI and providers are separate.", "sourceUrls": [ "https://zed.dev/blog/zed-is-now-open-source", "https://zed.dev/docs/ai/privacy-and-security" ], "confidence": "high", "unknowns": "This score covers the harness/editor source only; hosted services, models, data, or enterprise features may remain proprietary.", "suggestedScore": 9, "scoreReason": "Zed publishes its editor code under GPL-3.0 and its collaboration server under AGPL-3.0, making those components inspectable under their respective licenses. Its hosted AI features and selected models are separate services. Inspect the relevant component license before deciding how to modify or deploy it.", "argument": "Zed publishes its editor code under GPL-3.0 and its collaboration server under AGPL-3.0, making those components inspectable under their respective licenses. Its hosted AI features and selected models are separate services. Inspect the relevant component license before deciding how to modify or deploy it.", "assessmentType": "evidence", "score": 9, "href": "/harnesses/zed/#openness" }, "labour": { "summary": "Zed’s public code and opt-in training-data documentation provide visibility into contributions and collection choices, but not contributor compensation or worker conditions.", "sourceUrls": [ "https://github.com/zed-industries/zed", "https://zed.dev/docs/ai/ai-improvement" ], "confidence": "low", "unknowns": "Current worker pay, working conditions, subcontracting, safety provisions, and grievance outcomes are not established for this product’s full service chain.", "suggestedScore": null, "scoreReason": "Zed’s public source and opt-in training rules make some product development choices inspectable, but do not reveal contributor compensation or model-provider labor practices. The separate Zeta training dataset is not a workforce report. Support the project if you rely on it and review labor practices for the chosen external model.", "argument": "Zed’s public source and opt-in training rules make some product development choices inspectable, but do not reveal contributor compensation or model-provider labor practices. The separate Zeta training dataset is not a workforce report. Support the project if you rely on it and review labor practices for the chosen external model.", "assessmentType": "conditional", "score": null, "href": "/harnesses/zed/#labour" }, "military-contracts": { "summary": "No harness-specific defense deployment was established in the sources reviewed; provider and parent-company records need separate attribution.", "sourceUrls": [ "https://zed.dev/docs/ai/use-a-local-model" ], "confidence": "low", "unknowns": "No public evidence in the reviewed material names this harness in a defense workload; absence in this review does not establish absence of indirect use.", "suggestedScore": null, "scoreReason": "The product sources reviewed do not identify Zed as used in a named defense, weapons, or surveillance deployment. Where this harness lets users choose models or hosts, the underlying provider’s contracts and controls can change the picture; no relationship for this product is established here. Check the selected model, host, and company contract record before treating the tool as suitable for sensitive work.", "argument": "The product sources reviewed do not identify Zed as used in a named defense, weapons, or surveillance deployment. Where this harness lets users choose models or hosts, the underlying provider’s contracts and controls can change the picture; no relationship for this product is established here. Check the selected model, host, and company contract record before treating the tool as suitable for sensitive work.", "assessmentType": "conditional", "score": null, "href": "/harnesses/zed/#military-contracts" }, "israel-palestine": { "summary": "We did not find Zed Industries among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Zed Industries is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Check the model and host this tool connects to as well as the company providing the tool.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/harnesses/zed/#israel-palestine" } }, "sources": [ { "url": "https://zed.dev/blog/zed-is-now-open-source", "title": "Zed official privacy or open-source documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://zed.dev/docs/ai/privacy-and-security", "title": "Official privacy or data-processing disclosure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://zed.dev/docs/business/privacy", "title": "Official privacy or data-processing disclosure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://zed.dev/docs/ai/use-a-local-model", "title": "Zed local model configuration", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://zed.dev/docs/ai/ai-improvement", "title": "Zed AI improvement and opt-in training data", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://zed.dev/blog/zeta2", "title": "Zed Zeta2 training data and model disclosure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/zed-industries/zed", "title": "Zed", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/harnesses/zed/", "overall": { "score": 8.3, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "green", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 1 } }, { "id": "gpt-6", "name": "GPT-6 family", "category": "models", "organization": "OpenAI", "website": "https://openai.com/", "summary": "Closed, hosted frontier model family. GPT-6 Astra is the flagship; GPT-6.1 Sol and GPT-6 Luna are listed as lower-cost and efficient variants.", "scope": "The model identity and availability are release-specific; privacy, training-data sourcing, energy, labour, and customer relationships below are organization-level unless explicitly stated. Do not infer that a customer contract used this exact checkpoint.", "releaseNote": "Current official API catalog lists GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna. Checked 2026-10-23. The catalog does not expose a separate training-data or environmental model card for each variant.", "dimensions": { "energy": { "summary": "Disclosure gap: OpenAI has described data-center investments, but the sources checked do not give GPT-6 training energy, inference energy, water, or emissions per task. Hosted compute means user-side estimates need provider-region and workload data.", "sourceUrls": [ "https://openai.com/index/openai-joins-ports-pike-project/" ], "confidence": "low", "unknowns": "No GPT-6-specific energy, water, or lifecycle figures; announced infrastructure estimates are not model measurements.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The project announcement concerns planned data-center capacity, not a measured GPT-6 training run or representative request. A fair comparison needs model-level power, emissions, and workload boundaries; buyers can ask providers for region- and usage-specific reporting.", "score": null, "href": "/models/gpt-6/#energy" }, "training-rights": { "summary": "OpenAI says its models use public internet information, third-party partnerships, and information supplied or generated by users, human trainers, and researchers; it describes exclusions and opt-outs. This is a general account, not a GPT-6 dataset manifest or independent rights audit.", "sourceUrls": [ "https://openai.com/index/approach-to-data-and-ai/", "https://openai.com/policies/how-chatgpt-and-our-foundation-models-are-developed/" ], "confidence": "medium", "unknowns": "No GPT-6 source-level corpus list, rights clearance audit, or complete opt-out coverage was found.", "suggestedScore": 4, "scoreReason": "Broad data-source and opt-out statements exist, but no GPT-6-specific corpus or independent rights audit.", "assessmentType": "conditional", "argument": "OpenAI describes several data-source classes and opt-out or exclusion practices, but does not identify which sources trained GPT-6 or audit those rights independently. These policies show some process transparency, but customers seeking provenance still need dataset-level attestations.", "score": 4, "href": "/models/gpt-6/#training-rights" }, "privacy": { "summary": "Consumer ChatGPT content may be used to improve models unless users opt out; business and API data are not used for training by default. This describes service controls, not a guarantee about every downstream host.", "sourceUrls": [ "https://openai.com/policies/how-your-data-is-used-to-improve-model-performance/", "https://openai.com/business-data/", "https://developers.openai.com/api/docs/guides/your-data" ], "confidence": "high", "unknowns": "Retention, human review, and jurisdiction depend on product, account settings, and endpoint.", "suggestedScore": 6, "scoreReason": "Published opt-out for consumer data and default exclusion for business/API data, with product-dependent retention.", "assessmentType": "conditional", "argument": "The cited policies distinguish consumer ChatGPT from business/API use: consumer conversations have opt-out controls, while business/API inputs are excluded from training by default. Retention and access still vary by endpoint and settings, so compare the exact product contract and controls before sending sensitive data.", "score": 6, "href": "/models/gpt-6/#privacy" }, "openness": { "summary": "Closed model weights and proprietary training pipeline; access is through OpenAI products and API/partners. Public model names and API documentation do not make the model open-weight or reproducible.", "sourceUrls": [ "https://developers.openai.com/api/docs/models" ], "confidence": "high", "unknowns": "No public weights, full training data, or complete training recipe found.", "suggestedScore": 2, "scoreReason": "Closed API/product access with no public weights or reproducible training recipe.", "assessmentType": "evidence", "argument": "The official catalog offers GPT-6 as a product/API service but provides no downloadable weights or reproducible training materials. Users cannot independently inspect the checkpoint or host it themselves.", "score": 2, "href": "/models/gpt-6/#openness" }, "labour": { "summary": "Company context: OpenAI. OpenAI’s supplier code sets broad labor standards but does not publish worker outcomes. TIME’s investigation reviewed OpenAI–Sama contracts, payslips, and interviews about 2021–22 Kenyan toxic-content labeling; workers described low take-home pay and distress, while Sama disputed workload/counseling claims and OpenAI said Sama managed pay and support. TIME later reported OpenAI said it was unaware of the conditions and no longer worked with Sama; this is a historical contractor case, not evidence of current API annotation conditions. This is a record of the company behind GPT-6 family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://openai.com/policies/supplier-code/", "https://time.com/6247678/openai-chatgpt-kenya-workers/", "https://time.com/7012787/mophat-okinyi/", "https://openai.com/index/approach-to-data-and-ai/", "https://openai.com/index/learning-to-summarize-with-human-feedback/" ], "confidence": "medium", "unknowns": "Working conditions for the people building or supporting GPT-6 family are not established by this company-wide record. The sources do not establish whether OpenAI uses other current annotation vendors, present worker outcomes, or the effectiveness of the supplier code across vendors.", "suggestedScore": 4, "scoreReason": "Company-level assessment: A documented past supplier episode includes worker reports of low pay and psychological distress, countered by Sama and OpenAI responses; the current supplier code has no public independent outcome audit. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "A documented past supplier episode includes worker reports of low pay and psychological distress, countered by Sama and OpenAI responses; the current supplier code has no public independent outcome audit. This is relevant when deciding whose business to support, while pay and working conditions for the team behind GPT-6 family remain unestablished. Ask for current worker outcomes for that team and its suppliers.", "assessmentScope": "Company record", "productArgument": "OpenAI’s general materials acknowledge human trainers and feedback, but they do not report GPT-6 worker numbers, compensation, vendors, or working conditions. Because the evidence confirms human contribution but not its terms, seek supplier disclosures rather than treating silence as evidence of good or bad practice.", "score": 4, "href": "/models/gpt-6/#labour" }, "military-contracts": { "summary": "Company context: OpenAI. OpenAI publicly describes a 2026 classified Department of War agreement with cloud-only deployment, retained safety controls, cleared staff, and contractual limits. It separately says ChatGPT will reach GenAI.mil for unclassified work used by 3 million civilian and military personnel. Both are direct defense deployments; limits are company-reported and not independently audited. This is a record of the company behind GPT-6 family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://openai.com/index/our-agreement-with-the-department-of-war/", "https://openai.com/index/bringing-chatgpt-to-genaimil/" ], "confidence": "high", "unknowns": "Deployment of GPT-6 family under these contracts is not established. Actual use, oversight effectiveness, and compliance with contract limits are not independently established.", "suggestedScore": 5, "scoreReason": "Company-level assessment: OpenAI reports cloud-only deployment, retained safety controls, cleared staff, and contractual limits; these are meaningful safeguards, but no independent check of enforcement is public. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "OpenAI reports cloud-only deployment, retained safety controls, cleared staff, and contractual limits; these are meaningful safeguards, but no independent check of enforcement is public. This is relevant when deciding whose business to support. Deployment of GPT-6 family or this particular model release under those contracts remains unestablished; consider the company relationship separately from the product’s actual use.", "assessmentScope": "Company record", "productArgument": "OpenAI describes classified Department of War work and a GenAI.mil deployment, but neither public announcement attributes each service to GPT-6 specifically. Treat this as a company-level defense relationship, not proof of this checkpoint’s use; buyers with restrictions should obtain model-and-workflow deployment assurances.", "score": 5, "href": "/models/gpt-6/#military-contracts" }, "israel-palestine": { "summary": "We did not find OpenAI among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "OpenAI is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. OpenAI’s April 2026 agreement names Microsoft as its primary cloud partner and a major shareholder, with revenue-share payments to Microsoft through 2030. That is a documented commercial link to a BDS target, not a separate official listing of OpenAI; local use of downloaded weights does not by itself establish a new payment to either company.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://openai.com/index/next-phase-of-microsoft-partnership/" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": { "label": "Microsoft partnership", "sourceIds": [ "openai-ms" ], "argument": "OpenAI’s April 2026 agreement names Microsoft as its primary cloud partner and a major shareholder, with revenue-share payments to Microsoft through 2030. That is a documented commercial link to a BDS target, not a separate official listing of OpenAI; local use of downloaded weights does not by itself establish a new payment to either company." }, "businessContext": null, "score": null, "href": "/models/gpt-6/#israel-palestine" } }, "sources": [ { "url": "https://developers.openai.com/api/docs/models", "title": "Models | OpenAI API", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/approach-to-data-and-ai/", "title": "Our approach to data and AI", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/policies/how-your-data-is-used-to-improve-model-performance/", "title": "How your data is used to improve model performance", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/business-data/", "title": "Business data privacy, security, and compliance", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/our-agreement-with-the-department-of-war/", "title": "Our agreement with the Department of War", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/bringing-chatgpt-to-genaimil/", "title": "Bringing ChatGPT to GenAI.mil", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/openai-joins-ports-pike-project/", "title": "OpenAI joins PORTS-Pike project", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/learning-to-summarize-with-human-feedback/", "title": "Learning to summarize with human feedback", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/policies/how-chatgpt-and-our-foundation-models-are-developed/", "title": "How ChatGPT and our foundation models are developed", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://developers.openai.com/api/docs/guides/your-data", "title": "Data controls in the OpenAI platform", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/policies/supplier-code/", "title": "OpenAI Supplier Code of Conduct", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://time.com/6247678/openai-chatgpt-kenya-workers/", "title": "OpenAI used Kenyan workers on less than $2 per hour: TIME investigation", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://time.com/7012787/mophat-okinyi/", "title": "Mophat Okinyi: TIME100 AI 2024", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/next-phase-of-microsoft-partnership/", "title": "OpenAI: amended Microsoft partnership, April 27 2026", "kind": "primary", "checkedAt": "2026-10-23" } ], "href": "/models/gpt-6/", "overall": { "score": 4.2, "rated": 5, "total": 6, "coverage": 0.8333333333333334, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 13 } }, { "id": "claude-5", "name": "Claude 5.5 family", "category": "models", "organization": "Anthropic", "website": "https://www.anthropic.com/claude", "summary": "Closed, hosted model family. Claude Opus 5.5 is the latest verified Opus release; Anthropic's October model report also lists Sonnet 5.5 and Haiku 5.5.", "scope": "Release and training-data details come from Anthropic's family model report; privacy and customer relationships describe Anthropic products/company, not every Claude host or exact checkpoint.", "releaseNote": "Anthropic's Transparency Hub lists Claude Sonnet 5.5, Haiku 5.5, and Opus 5.5, with Opus 5.5 announced September 22, 2026. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "Disclosure gap: the public Opus 5.5 model report describes training and safeguards but no model-specific training energy, inference energy, water, or emissions. Anthropic's efficiency/cost claims are not energy measurements.", "sourceUrls": [ "https://www.anthropic.com/transparency" ], "confidence": "low", "unknowns": "No comparable per-query or training-run environmental figures found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Anthropic’s transparency materials discuss model development but provide no Opus 5.5 training-energy or comparable inference measurement. The model’s compute footprint cannot be compared yet; request model- and region-specific reporting before describing it as low-impact.", "score": null, "href": "/models/claude-5/#energy" }, "training-rights": { "summary": "Anthropic says the model was trained on a proprietary mix of public internet information, public and private datasets, synthetic data, and in some cases user data provided or permitted for training. This is a high-level disclosure, not an itemized rights ledger.", "sourceUrls": [ "https://www.anthropic.com/transparency" ], "confidence": "medium", "unknowns": "Exact dataset provenance, licensing and contested works are not enumerated in the model report.", "suggestedScore": 4, "scoreReason": "Training categories are disclosed, but sources and rights are not itemized.", "assessmentType": "conditional", "argument": "Anthropic describes a proprietary mix that can include public internet, licensed/private sources, synthetic data, and some user-provided data, but does not expose a Claude 5.5 source ledger. The breadth of disclosure is better than no account, yet rights and opt-out coverage cannot be verified corpus by corpus.", "score": 4, "href": "/models/claude-5/#training-rights" }, "privacy": { "summary": "Consumer chats may be used to improve Claude when the user permits it, for flagged safety review, or when feedback is submitted; commercial services are governed separately. Data retention and access vary by surface and covered-model status.", "sourceUrls": [ "https://privacy.anthropic.com/en/articles/10023555-how-do-you-use-personal-data-in-model-training", "https://support.claude.com/en/articles/15425996-data-retention-practices-for-covered-models" ], "confidence": "high", "unknowns": "Review policy and retention depend on product, settings, and model classification.", "suggestedScore": 6, "scoreReason": "Consumer training controls and distinct commercial terms are documented, with safety-review and retention exceptions.", "assessmentType": "conditional", "argument": "Anthropic’s policy differentiates consumer conversations and commercial services, and its retention page adds product-specific time limits for covered models. Since training use and retention depend on consent, feedback, safety flags, and service tier, use the exact account terms and settings for the intended workload.", "score": 6, "href": "/models/claude-5/#privacy" }, "openness": { "summary": "Claude 5 models are proprietary hosted models; the report describes a proprietary data mixture and distribution through Anthropic and partners, with no public weights or reproducible training pipeline.", "sourceUrls": [ "https://www.anthropic.com/transparency" ], "confidence": "high", "unknowns": "No public model weights or complete training data/recipe found.", "suggestedScore": 2, "scoreReason": "Proprietary hosted model; no public weights or complete training recipe.", "assessmentType": "evidence", "argument": "Anthropic’s transparency report describes proprietary training and distribution without publishing Claude 5.5 weights. Safety documentation is available, but users cannot independently inspect or host the model.", "score": 2, "href": "/models/claude-5/#openness" }, "labour": { "summary": "Company context: Anthropic. Anthropic’s 2025 Claude 4 system card and inbound-services agreement disclose use of paid data-labeling contractors and require crowd-work vendors to follow minimum worker-wellbeing expectations. Stanford FMTI found no public worker compensation, location, or top-source labor-practice outcomes, so the standards remain policy commitments rather than independently verified results. This is a record of the company behind Claude 5.5 family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://www-cdn.anthropic.com/de8ba9b01c9ab7cbabf5c33b80b7bbc618857627/Model%5C_Card%5C_Claude%5C_3.pdf", "https://crfm.stanford.edu/fmti/December-2025/company-reports/Anthropic_FinalReport_FMTI2025.html", "https://www.anthropic.com/legal/inbound-services-agreement", "https://www.anthropic.com/news/claudes-constitution", "https://www.anthropic.com/transparency" ], "confidence": "medium", "unknowns": "Working conditions for the people building or supporting Claude 5.5 family are not established by this company-wide record. Current pay, vendor locations, worker exposure, and compliance/audit results are not published per major data source.", "suggestedScore": 4, "scoreReason": "Company-level assessment: Public vendor-wellbeing standards and contractor disclosure are meaningful, but Stanford FMTI found no compensation or worker-level outcomes for the principal data sources. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "conditional", "argument": "Public vendor-wellbeing standards and contractor disclosure are meaningful, but Stanford FMTI found no compensation or worker-level outcomes for the principal data sources. This is relevant when deciding whose business to support, while pay and working conditions for the team behind Claude 5.5 family remain unestablished. Ask for current worker outcomes for that team and its suppliers.", "assessmentScope": "Company record", "productArgument": "Anthropic has published research describing contractors in preference-feedback work, while its current model report does not disclose Opus 5.5 workforce scale or conditions. That shows human labor is part of the development process but leaves the treatment of workers unassessed; seek vendor and pay disclosures.", "score": 4, "href": "/models/claude-5/#labour" }, "military-contracts": { "summary": "Company context: Anthropic. Anthropic reports a two-year DoD CDAO prototype agreement with a $200 million ceiling, Claude in classified Palantir mission workflows, and Claude Gov deployments on AWS for national-security agencies. In March 2026, Anthropic said DoW designated it a supply-chain risk, narrowly applying to Claude use directly under DoW contracts, and that it would support a transition for warfighters. This is a record of the company behind Claude 5.5 family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://www.anthropic.com/news/anthropic-and-the-department-of-defense-to-advance-responsible-ai-in-defense-operations", "https://www.anthropic.com/news/where-stand-department-war" ], "confidence": "high", "unknowns": "Deployment of Claude 5.5 family under these contracts is not established. These company statements do not establish the specific operational outcomes or independently assess the safeguards of deployments.", "suggestedScore": 3, "scoreReason": "Company-level assessment: The company documents classified defense use and a $200 million prototype agreement; deployments and guardrails are not independently assessed. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "The company documents classified defense use and a $200 million prototype agreement; deployments and guardrails are not independently assessed. This is relevant when deciding whose business to support. Deployment of Claude 5.5 family or this particular model release under those contracts remains unestablished; consider the company relationship separately from the product’s actual use.", "assessmentScope": "Company record", "productArgument": "Anthropic publicly announced a time-limited Defense Department prototype agreement and described Claude in classified Palantir workflows. The announcement is evidence of a company relationship, but not a complete model/version deployment list; users should treat military use as a documented provider relationship and verify the exact endpoint.", "score": 3, "href": "/models/claude-5/#military-contracts" }, "israel-palestine": { "summary": "We did not find Anthropic among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Anthropic is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Anthropic names AWS as its primary training and cloud provider and reports a large compute commitment and Amazon investment. This connects its business to a BDS pressure target without making Anthropic a separately named target; the announcement does not identify the host of every Claude request.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://www.anthropic.com/news/anthropic-amazon-compute" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": { "label": "Amazon partnership", "sourceIds": [ "anthropic-aws" ], "argument": "Anthropic names AWS as its primary training and cloud provider and reports a large compute commitment and Amazon investment. This connects its business to a BDS pressure target without making Anthropic a separately named target; the announcement does not identify the host of every Claude request." }, "businessContext": null, "score": null, "href": "/models/claude-5/#israel-palestine" } }, "sources": [ { "url": "https://www.anthropic.com/claude-opus-5-5", "title": "Claude Opus 5.5 | Anthropic", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/transparency", "title": "Anthropic’s Transparency Hub | Anthropic", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://privacy.anthropic.com/en/articles/10023555-how-do-you-use-personal-data-in-model-training", "title": "How Do You Use Personal Data in Model Training?", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://support.claude.com/en/articles/15425996-data-retention-practices-for-covered-models", "title": "Data retention practices for Covered Models", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/news/claudes-constitution", "title": "Claude's constitution", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/news/anthropic-and-the-department-of-defense-to-advance-responsible-ai-in-defense-operations", "title": "Anthropic and the Department of Defense to advance responsible AI in defense operations", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www-cdn.anthropic.com/de8ba9b01c9ab7cbabf5c33b80b7bbc618857627/Model%5C_Card%5C_Claude%5C_3.pdf", "title": "Claude 3 Model Card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://crfm.stanford.edu/fmti/December-2025/company-reports/Anthropic_FinalReport_FMTI2025.html", "title": "Stanford FMTI 2025 Anthropic transparency assessment", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/legal/inbound-services-agreement", "title": "Anthropic Inbound Services Agreement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/news/where-stand-department-war", "title": "Where things stand with the Department of War", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/news/anthropic-amazon-compute", "title": "Anthropic: expanded Amazon compute agreement, April 2026", "kind": "primary", "checkedAt": "2026-10-23" } ], "href": "/models/claude-5/", "overall": { "score": 3.8, "rated": 5, "total": 6, "coverage": 0.8333333333333334, "partial": true, "color": "red", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 16 } }, { "id": "gemini-4-argon", "name": "Gemini 4 Argon", "category": "models", "organization": "Google DeepMind / Google", "website": "https://deepmind.google/models/gemini/", "summary": "Hosted frontier model released first to defenders and then rolled out broadly in October 2026.", "scope": "Argon model-card statements are model-specific. Google privacy practices and Project Nimbus are company/service-level findings; neither establishes a particular use of Argon by an Israeli agency.", "releaseNote": "Google announced Gemini 4 Argon on September 30 and broad availability on October 13, 2026. Its October model card identifies it as a new, non-fine-tuned frontier model. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "Argon was trained on Google TPUs; its card says TPU efficiency aligns with Google's sustainability commitments. It publishes no Argon training energy, inference energy, water, or emissions totals, so this is a hardware/process disclosure rather than an impact measurement.", "sourceUrls": [ "https://deepmind.google/models/model-cards/gemini-4-argon/" ], "confidence": "medium", "unknowns": "No model-run energy, water or carbon data, and no per-query estimate.", "suggestedScore": null, "scoreReason": "No score: the available sources report hardware, throughput, or provider-wide efficiency framing, but no model-specific energy, water, or emissions impact. Efficiency claims alone do not establish a lower footprint.", "assessmentType": "conditional", "argument": "The Argon card names TPU training and offers only a general efficiency/sustainability statement, with no run-level energy or emissions figures. Hardware choice alone cannot establish footprint; a model-bounded energy and inference report is still needed.", "score": null, "href": "/models/gemini-4-argon/#energy" }, "training-rights": { "summary": "Argon's card identifies public web data, crawled data, licensed commercial data, user data under applicable terms and controls, workforce/business data, and synthetic data; it says filtering includes robots.txt. The card does not list datasets or license-level coverage.", "sourceUrls": [ "https://deepmind.google/models/model-cards/gemini-4-argon/" ], "confidence": "high", "unknowns": "No itemized corpus, dataset-level rights audit, or independent assessment.", "suggestedScore": 5, "scoreReason": "Model card discloses several data classes and robots.txt filtering, but not an itemized rights audit.", "assessmentType": "conditional", "argument": "The card names public web, licensed data, controlled user data, workforce/business data, and synthetic data. It does not map any category to specific examples or establish permissions, so the source description is informative but not a rights audit.", "score": 5, "href": "/models/gemini-4-argon/#training-rights" }, "privacy": { "summary": "Google states that a subset of consumer Gemini chats can be human-reviewed and retained up to three years; the user can manage activity settings. Enterprise/API terms differ. This concerns Gemini Apps consumer interactions, not all Argon endpoints.", "sourceUrls": [ "https://support.google.com/gemini/answer/13594961" ], "confidence": "high", "unknowns": "Exact behavior depends on product and account controls; no complete cross-surface retention map was assessed.", "suggestedScore": 4, "scoreReason": "Consumer chats may receive human review and retention up to three years; enterprise/API controls differ.", "assessmentType": "evidence", "argument": "Google’s consumer Gemini policy says a subset of chats may be human-reviewed and retained for up to three years, while activity controls affect handling; the model card itself does not define this service policy. Avoid putting sensitive consumer prompts into that service, and assess enterprise/API terms separately.", "score": 4, "href": "/models/gemini-4-argon/#privacy" }, "openness": { "summary": "Argon is a hosted model distributed through Gemini products and API; Google publishes a model card but no weights or full training recipe. A model card is transparency documentation, not open-weight access.", "sourceUrls": [ "https://deepmind.google/models/model-cards/gemini-4-argon/", "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/" ], "confidence": "high", "unknowns": "No public weights or full reproducibility artifacts found.", "suggestedScore": 2, "scoreReason": "Hosted model with model card but no weights or full training artifacts.", "assessmentType": "evidence", "argument": "Argon has an official model card and product/API distribution but no public weights or complete training recipe. Documentation earns transparency credit, while actual inspection, self-hosting, and reproduction remain constrained.", "score": 2, "href": "/models/gemini-4-argon/#openness" }, "labour": { "summary": "Company context: Google DeepMind / Google. Google’s supplier code applies labor and human-rights requirements across direct, temporary, migrant, contract, and subcontracted workers; it allows audits, requires remediation, and sets a 60-hour weekly limit with stated exceptions. This is a supplier policy, not an audit of AI data workers. This is a record of the company behind Gemini 4 Argon; it does not establish use of this particular product or release.", "sourceUrls": [ "https://about.google/company-info/supplier-code-of-conduct/", "https://deepmind.google/models/model-cards/gemini-4-argon/" ], "confidence": "high", "unknowns": "Working conditions for the people building or supporting Gemini 4 Argon are not established by this company-wide record. No current AI-annotation-specific outcomes, public audit findings, or remediation data for Vertex-related data work were established.", "suggestedScore": 4, "scoreReason": "Company-level assessment: Google publishes broad supplier protections and audit rights, but this review found no worker-level results for AI data work. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "Google publishes broad supplier protections and audit rights, but this review found no worker-level results for AI data work. This is relevant when deciding whose business to support, while pay and working conditions for the team behind Gemini 4 Argon remain unestablished. Ask for current worker outcomes for that team and its suppliers.", "assessmentScope": "Company record", "productArgument": "The Argon card confirms human-preference data and workforce data were used, but does not tell who performed annotation, where, or under what terms. This establishes human involvement without evidence of labor safeguards; request current contractor conditions before drawing a labor conclusion.", "score": 4, "href": "/models/gemini-4-argon/#labour" }, "military-contracts": { "summary": "Company context: Google DeepMind / Google. Google documents access for Department of War personnel and classified defense workflows. Separately, Israel’s procurement administration identifies Google as a Nimbus public-cloud provider; AFSC and Who Profits describe military/security use, while the Guardian reports contract terms limiting service withdrawal. These sources do not establish Google’s role in a particular attack. This is a record of the company behind Gemini 4 Argon; it does not establish use of this particular product or release.", "sourceUrls": [ "https://cloud.google.com/blog/topics/public-sector/announcements-and-highlights-from-google-public-sector-summit-2026", "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/alphabet", "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/", "https://blog.google/innovation-and-ai/technology/safety-security/gemini-4-model-safety-security/" ], "confidence": "high", "unknowns": "Deployment of Gemini 4 Argon under these contracts is not established. The source does not disclose contract terms or evaluate downstream deployment outcomes.", "suggestedScore": 3, "scoreReason": "Company-level assessment: Google documents U.S. Department of War and classified defense workflows; Israel’s procurement record identifies Google as a Nimbus cloud provider, and independent research/reporting describes military or security use and limits on withdrawal. The reviewed evidence does not attribute a specific harmful operation to Google. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "conditional", "argument": "Google documents U.S. Department of War and classified defense workflows; Israel’s procurement record identifies Google as a Nimbus cloud provider, and independent research/reporting describes military or security use and limits on withdrawal. The reviewed evidence does not attribute a specific harmful operation to Google. This is relevant when deciding whose business to support. Deployment of Gemini 4 Argon or this particular model release under those contracts remains unestablished; consider the company relationship separately from the product’s actual use.", "assessmentScope": "Company record", "productArgument": "Google’s launch materials describe Fairwind access for trusted cyber defenders, which is a security partnership rather than evidence of military deployment of Argon. Fairwind is a cyber-defender partnership; keep it distinct from corporate defense ties and do not describe it as a military contract without further evidence.", "score": 3, "href": "/models/gemini-4-argon/#military-contracts" }, "israel-palestine": { "summary": "Google / Alphabet is named as a pressure target in the BNC’s official campaign sources. The product is developed or supplied by Google / Alphabet.", "argument": "If following this BNC campaign is a requirement, consider alternatives to paid Google / Alphabet products where practical. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well. This label follows the named company’s campaign status; it does not claim that BDS separately reviewed this exact product.", "assessmentType": "evidence", "confidence": "high", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "pressure", "statusLabel": "Pressure target", "reviewedAt": "2026-10-23", "assessmentScope": "Developer / service company", "bdsMatch": { "target": "Google / Alphabet", "status": "pressure", "scope": "Developer / service company", "relationship": "The product is developed or supplied by Google / Alphabet.", "application": "If following this BNC campaign is a requirement, consider alternatives to paid Google / Alphabet products where practical.", "sourceIds": [] }, "upstreamExposure": null, "businessContext": { "summary": "Company context: Google DeepMind / Google. Israel’s procurement administration says Google was selected with AWS for Nimbus public-cloud services and lists the Ministry of Defense and IDF among project partners. The Guardian’s investigation of leaked agreement documents reports that Nimbus terms limited Google’s ability to suspend service, including after alleged policy violations. Who Profits and AFSC describe cloud and AI use by military/security bodies, including surveillance-related capabilities; these are attributed research claims, not proof of Google’s role in a specific attack. Both companies denied circumventing legal obligations in response to the Guardian investigation. This is a record of the company behind Gemini 4 Argon; it does not establish use of this particular product or release.", "unknowns": "How Gemini 4 Argon is used in this relationship is not established. The government record establishes the contract and project participants; the investigative reporting describes contract controls, and research organizations make broader use claims. The reviewed evidence does not establish Google’s contribution to a particular operation, injury, or other specific harm.", "sourceUrls": [ "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/alphabet", "https://cloud.google.com/blog/topics/inside-google-cloud/google-cloud-selected-to-provide-cloud-services-to-the-state-of-israel" ] }, "score": null, "href": "/models/gemini-4-argon/#israel-palestine" } }, "sources": [ { "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/", "title": "Introducing Gemini 4 Argon", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/", "title": "Gemini 4 Argon is now available for everyone", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://deepmind.google/models/model-cards/gemini-4-argon/", "title": "Gemini 4 Argon - Model Card | Google DeepMind", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://support.google.com/gemini/answer/13594961", "title": "Gemini Apps Privacy Hub", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cloud.google.com/blog/topics/inside-google-cloud/google-cloud-selected-to-provide-cloud-services-to-the-state-of-israel", "title": "Google Cloud selected to provide cloud services to the State of Israel", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://blog.google/innovation-and-ai/technology/safety-security/gemini-4-model-safety-security/", "title": "Developing Gemini 4 Argon responsibly and safely", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://about.google/company-info/supplier-code-of-conduct/", "title": "Google Supplier Code of Conduct", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cloud.google.com/blog/topics/public-sector/announcements-and-highlights-from-google-public-sector-summit-2026", "title": "Announcements and highlights from Google Public Sector Summit 2026", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://mr.gov.il/ilgstorefront/en/news/details/111222", "title": "Nimbus Tender — Israel Government Procurement Administration", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "title": "Revealed: Israel demanded Google and Amazon use secret “wink” to sidestep legal orders", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://www.whoprofits.org/publications/report/164", "title": "Digitizing Occupation: The Role of Big Tech in Israeli Military Infrastructure", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://investigate.afsc.org/company/alphabet", "title": "Alphabet / Google — AFSC Investigate", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/gemini-4-argon/", "overall": { "score": 3.6, "rated": 5, "total": 6, "coverage": 0.8333333333333334, "partial": true, "color": "red", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 17 } }, { "id": "grok-4-7", "name": "Grok 4.7", "category": "models", "organization": "xAI", "website": "https://x.ai/", "summary": "Hosted frontier model for coding and knowledge work; the developer describes a new base model and extended reinforcement-learning run.", "scope": "Release notes are Grok 4.7-specific; privacy summaries concern consumer Grok and X, and should not be assumed to govern every API or partner deployment.", "releaseNote": "xAI announced Grok 4.7 on September 21, 2026 and lists it as the current flagship in its model catalog. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "Disclosure gap: xAI's release describes speed and price improvements but gives no Grok 4.7 training energy, inference energy, water, or carbon figures. Throughput or token-cost claims are not environmental measurements.", "sourceUrls": [ "https://x.ai/news/grok-4-7" ], "confidence": "low", "unknowns": "No model-specific compute or environmental report found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The Grok 4.7 release reports speed and price improvements, neither of which measures electricity, emissions, or water per task. A faster model may change total demand as well as per-task use; do not infer either direction without workload-level measurements.", "score": null, "href": "/models/grok-4-7/#energy" }, "training-rights": { "summary": "xAI says Grok pretraining primarily uses publicly available information, with quality filters and efforts to minimize personal data. Its 4.7 release identifies reinforcement learning but does not publish a dataset-level rights inventory.", "sourceUrls": [ "https://x.ai/legal/faq", "https://x.ai/news/grok-4-7" ], "confidence": "medium", "unknowns": "Source-level dataset, licenses, consent and rights-resolution process are not disclosed in the checked sources.", "suggestedScore": 4, "scoreReason": "Publicly available training information and filtering are stated, but rights and corpus detail are not listed.", "assessmentType": "conditional", "argument": "xAI says Grok pretraining mainly uses public information with filters and describes reinforcement-learning data, but gives no source-level corpus or rights audit for 4.7. These are process descriptions, not proof that particular training sources were licensed; users needing traceability should request provenance and opt-out details.", "score": 4, "href": "/models/grok-4-7/#training-rights" }, "privacy": { "summary": "Grok consumer content may be used for training unless users opt out; Private Chat is excluded from training and is deleted within 30 days. X separately says public posts and Grok interactions may be shared with xAI, with controls and exceptions. API terms are separate.", "sourceUrls": [ "https://x.ai/legal/faq", "https://help.x.com/en/using-x/about-grok", "https://x.ai/legal/privacy-policy" ], "confidence": "high", "unknowns": "Controls differ by logged-in status, geography, product surface, and feedback submission.", "suggestedScore": 4, "scoreReason": "Consumer training opt-out and Private Chat controls are documented, with surface/geography exceptions.", "assessmentType": "conditional", "argument": "The cited xAI and X notices say consumer Grok data may train models unless users opt out, while Private Chat has separate no-training and deletion terms. Those controls are specific to consumer surfaces and do not automatically govern API or third-party hosting; check the service-specific policy.", "score": 4, "href": "/models/grok-4-7/#privacy" }, "openness": { "summary": "Grok 4.7 is a proprietary hosted/API model. The release identifies model changes at a high level but publishes no weights or full training pipeline.", "sourceUrls": [ "https://x.ai/news/grok-4-7", "https://docs.x.ai/developers/models" ], "confidence": "high", "unknowns": "No public weights or full training recipes found.", "suggestedScore": 2, "scoreReason": "Proprietary hosted/API release without public weights.", "assessmentType": "evidence", "argument": "The 4.7 release and developer catalog provide hosted/API access but no downloadable weights or reproducible training pipeline. This supports a low openness judgment; provider access should not be described as open access.", "score": 2, "href": "/models/grok-4-7/#openness" }, "labour": { "summary": "X says Grok training data is curated by human AI Tutors and the release describes reinforcement learning. The checked sources do not state tutor/annotator employment conditions, pay, location, contractors, or grievance mechanisms.", "sourceUrls": [ "https://help.x.com/en/using-x/about-grok", "https://x.ai/news/grok-4-7" ], "confidence": "medium", "unknowns": "No worker-condition disclosure found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "xAI says human AI Tutors curate Grok training data and the release describes reinforcement learning, but neither source gives tutor compensation, location, contractor status, or conditions. Human involvement is documented, but this does not establish worker protections; request terms before drawing a labor conclusion.", "score": null, "href": "/models/grok-4-7/#labour" }, "military-contracts": { "summary": "xAI’s government page says its products were made available to US government customers in July 2025 and names Grok among the family used by government and national-security customers; a current government page describes SpaceXSI use of the same models in classified and unclassified contexts for intelligence analysis, operational planning, and defense logistics. These disclosures do not establish Grok 4.7 as the exact deployed checkpoint.", "sourceUrls": [ "https://x.ai/news/government", "https://x.ai/grok/government" ], "confidence": "high", "unknowns": "The checked pages do not provide a model-by-model contract or deployment register, nor do they identify Grok 4.7 in the described government workflows.", "suggestedScore": 3, "scoreReason": "xAI directly describes government and defense/intelligence deployments of its model family, though the exact current checkpoint is not specified.", "assessmentType": "conditional", "argument": "xAI’s government pages explicitly describe its model family in government, classified, intelligence-analysis, planning, and defense-logistics settings, without naming Grok 4.7 as the exact checkpoint. This merits a serious company/family-level concern, while procurement screening should still ask which version and deployment are in scope.", "score": 3, "href": "/models/grok-4-7/#military-contracts" }, "israel-palestine": { "summary": "We did not find xAI among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "xAI is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/grok-4-7/#israel-palestine" } }, "sources": [ { "url": "https://x.ai/news/grok-4-7", "title": "Introducing Grok 4.7 | SpaceXSI", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.x.ai/developers/models", "title": "Grok models and pricing", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://x.ai/legal/faq", "title": "Consumer FAQs", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://help.x.com/en/using-x/about-grok", "title": "About Grok", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://x.ai/legal/privacy-policy", "title": "SpaceXAI Privacy Policy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://x.ai/news/government", "title": "xAI government", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://x.ai/grok/government", "title": "Grok for government", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/grok-4-7/", "overall": { "score": 3.3, "rated": 4, "total": 6, "coverage": 0.6666666666666666, "partial": true, "color": "red", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 19 } }, { "id": "llama-4", "name": "Llama 4 family", "category": "models", "organization": "Meta", "website": "https://ai.meta.com/llama/", "summary": "Open-weight multimodal models under Meta's Community License, including Llama 4 Scout and Maverick. The weights are available, while terms and scale restrictions mean this is not unrestricted open-source software.", "scope": "Llama model release/license statements are model-family-specific. Meta platform data practices and public human-feedback descriptions are company/family context and not proof of every Llama 4 training input.", "releaseNote": "Meta's Llama documentation still lists Llama 4 Scout and Maverick as its newest Llama releases on 2026-10-23. Meta separately markets Muse models; those are outside the requested Llama family record.", "dimensions": { "energy": { "summary": "Meta's Llama 4 model card reports pretraining at 7.38M H100 GPU-hours (Scout 5.0M, Maverick 2.38M), 700 W peak capacity per device, and estimated 1,999 tonnes CO2e location-based emissions. It reports 0 market-based tonnes based on Meta's company-wide clean-electricity matching. These are Meta's reported estimates; GPU-hours and rated device power are not measured facility electricity. The figures exclude fine-tuning and do not report inference energy.", "sourceUrls": [ "https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct" ], "confidence": "high", "unknowns": "No measured facility kWh/PUE, water, post-training footprint, or standardized inference energy per task/token. The 0 market-based CO2e relies on Meta-wide renewable matching, not a model-specific zero-emissions measurement.", "suggestedScore": 4, "scoreReason": "A high-impact pretraining disclosure is unusually specific and renewable matching is reported, but location-based emissions are estimated at 1,999 tonnes and inference/post-training footprints remain undisclosed; score reflects both scale and reporting transparency.", "assessmentType": "evidence", "argument": "Meta’s Llama 4 card reports GPU-hours and location-based CO2e for pretraining, but excludes fine-tuning and does not measure inference; its zero market-based figure relies on company-wide clean-electricity matching. The disclosed estimate is useful but incomplete, so compare the location-based boundary and ask for downstream serving impacts.", "score": 4, "href": "/models/llama-4/#energy" }, "training-rights": { "summary": "Meta's Llama 4 release describes its training approach but does not provide a full source-level dataset/rightsholder manifest in the checked material. Meta's general AI disclosures say it uses public, licensed, and Meta-product information; those policies do not establish legal rights for each Llama 4 work.", "sourceUrls": [ "https://ai.meta.com/blog/llama-4-multimodal-intelligence/", "https://www.sec.gov/Archives/edgar/data/1326801/000162828026025532/meta-20260416.htm" ], "confidence": "medium", "unknowns": "No complete Llama 4 dataset list, rights audit, or independent consent review found.", "suggestedScore": 4, "scoreReason": "General data categories are disclosed but Llama 4 has no complete source-level rights inventory.", "assessmentType": "conditional", "argument": "The release plus Meta’s filing describe data practices and rights processes, but do not provide a complete Llama 4 source/rightsholder manifest. These policies provide some process disclosure, but do not establish that every training item was licensed or consented.", "score": 4, "href": "/models/llama-4/#training-rights" }, "privacy": { "summary": "Meta says public posts and interactions with its AI products can be used to develop AI; it says private messages with friends and family are not used unless shared with AI. This is platform-level policy context, not inference privacy of self-hosted Llama weights.", "sourceUrls": [ "https://www.sec.gov/Archives/edgar/data/1326801/000162828026025532/meta-20260416.htm" ], "confidence": "medium", "unknowns": "Local deployment shifts prompt handling to the operator; Meta's platform policy is not applicable to every downstream host.", "suggestedScore": 7, "scoreReason": "Weights can be self-hosted for operator-controlled prompt handling; this is a deployment option rather than a model guarantee.", "assessmentType": "conditional", "argument": "Meta’s filing distinguishes public interactions and AI interactions from private messages, which are generally not used unless a user shares them with AI. This is a policy boundary rather than a guarantee across every product or region; users should review settings and avoid sharing private material into AI features.", "score": 7, "href": "/models/llama-4/#privacy" }, "openness": { "summary": "Weights and model cards are downloadable, but access is governed by the Llama 4 Community License Agreement, including conditions and scale-related provisions. Training data and full end-to-end training pipeline are not fully released with the weights.", "sourceUrls": [ "https://ai.meta.com/resources/models-and-libraries/llama-downloads/", "https://ai.meta.com/llama/get-started/" ], "confidence": "high", "unknowns": "Review the specific license and acceptable-use terms before treating a use as permitted; no full pretraining corpus is provided.", "suggestedScore": 6, "scoreReason": "Weights are available, but a custom community license and incomplete training-data release limit openness.", "assessmentType": "conditional", "argument": "The downloadable weights and model card enable local inspection, but the community license carries conditions and the training corpus is not fully open. The family is more accessible than a hosted-only model, while restrictions and missing data limit full reproducibility.", "score": 6, "href": "/models/llama-4/#openness" }, "labour": { "summary": "Meta has described human feedback and human annotations for earlier Llama generations. The sources checked do not disclose Llama 4 annotator workforce conditions, compensation, vendor chain, or worker protections.", "sourceUrls": [ "https://ai.meta.com/blog/meta-llama-3-meta-ai-responsibility/", "https://ai.meta.com/blog/llama-4-multimodal-intelligence/" ], "confidence": "low", "unknowns": "Earlier-generation human-feedback descriptions cannot establish Llama 4 labour conditions.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Meta’s earlier Llama responsibility materials discuss human feedback and annotation, but the Llama 4 announcement does not disclose current workers’ pay, vendors, or conditions. Those older-generation sources do not establish Llama 4 labor practices; current supplier details are needed.", "score": null, "href": "/models/llama-4/#labour" }, "military-contracts": { "summary": "Meta announced in September 2025 that Llama models were being used by US military and intelligence stakeholders, including a Legion Intelligence collaboration with USSOCOM on SOFChat; it also named Lockheed Martin and Anduril among defense partners and described Llama use in military workflows. These are Llama-family/company disclosures and do not identify Llama 4 as the exact deployed checkpoint.", "sourceUrls": [ "https://about.fb.com/news/2025/09/meta-supporting-us-national-security-with-ai/", "https://about.fb.com/news/2025/09/strengthening-us-national-security-by-making-llama-available-to-key-allies/" ], "confidence": "high", "unknowns": "The announcements are dated 2025 and do not give a full deployment inventory or identify exact model versions for each named use.", "suggestedScore": 3, "scoreReason": "Meta publicly describes real military and intelligence use and named defense-industry partners at the Llama-family level; exact Llama 4 deployment remains unverified.", "assessmentType": "evidence", "argument": "Meta’s 2025 announcements describe Llama use by military/intelligence stakeholders and named defense partners, including a USSOCOM-linked tool, while not identifying the exact checkpoint. That is a documented family-level relationship and a serious concern; a buyer should verify whether Llama 4 is used in its military workflow.", "score": 3, "href": "/models/llama-4/#military-contracts" }, "israel-palestine": { "summary": "Meta is named as a pressure target in the BNC’s official campaign sources. The product is developed or supplied by Meta.", "argument": "If following this BNC campaign is a requirement, consider alternatives to paid Meta products where practical. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well. This label follows the named company’s campaign status; it does not claim that BDS separately reviewed this exact product.", "assessmentType": "evidence", "confidence": "high", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "pressure", "statusLabel": "Pressure target", "reviewedAt": "2026-10-23", "assessmentScope": "Developer / service company", "bdsMatch": { "target": "Meta", "status": "pressure", "scope": "Developer / service company", "relationship": "The product is developed or supplied by Meta.", "application": "If following this BNC campaign is a requirement, consider alternatives to paid Meta products where practical.", "sourceIds": [] }, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/llama-4/#israel-palestine" } }, "sources": [ { "url": "https://ai.meta.com/llama/get-started/", "title": "Get started with Llama", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://ai.meta.com/resources/models-and-libraries/llama-downloads/", "title": "Llama model downloads and license information", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://ai.meta.com/blog/llama-4-multimodal-intelligence/", "title": "The Llama 4 herd", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://ai.meta.com/blog/meta-llama-3-meta-ai-responsibility/", "title": "Our responsible approach to Meta AI and Meta Llama 3", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.sec.gov/Archives/edgar/data/1326801/000162828026025532/meta-20260416.htm", "title": "Meta 2026 filing on AI data practices", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct", "title": "Llama 4 Scout model card: training energy and emissions", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://about.fb.com/news/2025/09/meta-supporting-us-national-security-with-ai/", "title": "Meta: Supporting US national security with AI", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://about.fb.com/news/2025/09/strengthening-us-national-security-by-making-llama-available-to-key-allies/", "title": "Meta: Strengthening US national security by making Llama available to key allies", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/llama-4/", "overall": { "score": 4.8, "rated": 5, "total": 6, "coverage": 0.8333333333333334, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 12 } }, { "id": "deepseek-v4-1", "name": "DeepSeek V4 family", "category": "models", "organization": "DeepSeek", "website": "https://www.deepseek.com/", "summary": "The latest verified API release is DeepSeek V4.1 Flash, updated October 17, 2026; DeepSeek V4 Pro remains available per the changelog.", "scope": "Release availability is from the official API changelog. The requested social dimensions are only assessed where a primary disclosure was located; absence of public detail is recorded as a gap.", "releaseNote": "Official API changelog says V4.1 Flash was upgraded to Flash-1017 on October 17, 2026 and V4 Pro would continue to be served. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "Disclosure gap: no V4.1 model-specific energy, water, or emissions data found. Efficiency and pricing claims cannot stand in for energy measurement.", "sourceUrls": [ "https://api-docs.deepseek.com/updates/" ], "confidence": "low", "unknowns": "No training compute or inference energy report in checked first-party material.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The API update log documents product versions but does not quantify V4.1 training or inference energy. Capability and pricing claims cannot fill that gap; a comparison needs measured training compute and serving data.", "score": null, "href": "/models/deepseek-v4-1/#energy" }, "training-rights": { "summary": "DeepSeek’s public V4 training-content summary (last updated April 8, 2026, and identifying V4-Pro/V4-Flash) reports more than 10 trillion text tokens; names Common Crawl, Stack Exchange, broad public corpora, commercially licensed data, and non-public third-party datasets; and describes DeepSeekBot collection with robots.txt/rights-reservation measures. It also marks user-interaction data and synthetic data as used, with de-identification and opt-out language. The summary does not identify every source or establish that all these details apply to the later V4.1-Flash-1017 update.", "sourceUrls": [ "https://cdn.deepseek.com/policies/DeepSeek-V4-public-summary-of-training-content.pdf", "https://cdn.deepseek.com/policies/en-US/model-algorithm-disclosure.html", "https://api-docs.deepseek.com/updates/" ], "confidence": "medium", "unknowns": "The April 2026 V4 summary is not a V4.1-Flash-1017-specific data ledger; exact update lineage, dataset-level licensing, and independent rights verification remain unavailable.", "suggestedScore": 4, "scoreReason": "The V4 disclosure identifies source categories, named corpora, licensing, and crawl controls, but exact V4.1-Flash-1017 coverage and independent rights clearance are not shown.", "assessmentType": "conditional", "argument": "DeepSeek’s V4 summary names major source categories and datasets, commercial licensing, a crawler that respects rights reservations, and user-data opt-out provisions. Because the report identifies V4-Pro/V4-Flash and predates V4.1-Flash-1017, it is useful family context rather than proof of the newer checkpoint’s complete provenance.", "score": 4, "href": "/models/deepseek-v4-1/#training-rights" }, "privacy": { "summary": "The official Responses API reference says responses and conversations are not stored on that endpoint; the October 17 changelog maps `deepseek-flash` to V4.1-Flash-1017. This is an endpoint-specific no-conversation-storage statement, not a promise that no operational telemetry or other API data are retained. DeepSeek’s separate first-party app policy describes collection and model-improvement use with opt-out, and excludes downstream developer-run apps.", "sourceUrls": [ "https://api-docs.deepseek.com/api/create-response/", "https://api-docs.deepseek.com/updates/", "https://cdn.deepseek.com/policies/en-US/deepseek-privacy-policy.html" ], "confidence": "high", "unknowns": "The stateless guarantee covers response/conversation storage for the documented endpoint; API telemetry, other API endpoints, and app-service handling require separate policy checks.", "suggestedScore": 4, "scoreReason": "The specific endpoint promises no server-stored conversations, while the broader first-party app policy permits collection and model-improvement use and does not settle API telemetry.", "assessmentType": "conditional", "argument": "DeepSeek documents no server-side response or conversation storage for its Responses endpoint, and its changelog connects `deepseek-flash` to V4.1-Flash-1017. That is a narrow retention safeguard, not a general no-logging promise; verify telemetry and other endpoint terms before sending sensitive prompts.", "score": 4, "href": "/models/deepseek-v4-1/#privacy" }, "openness": { "summary": "DeepSeek-V4.1-Flash-1017 is publicly downloadable from DeepSeek’s official model repository under the MIT License, with inference instructions and supporting code. The weights can be run locally, but the complete training corpus and end-to-end training pipeline are not fully public.", "sourceUrls": [ "https://api-docs.deepseek.com/news/news261017/", "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017", "https://cdn.deepseek.com/policies/en-US/model-algorithm-disclosure.html" ], "confidence": "high", "unknowns": "The release provides weights and inference materials but not a complete source-level corpus or end-to-end reproducible training process.", "suggestedScore": 8, "scoreReason": "Official MIT-licensed weights and inference artifacts enable broad local use and inspection; the undisclosed full corpus and training pipeline limit reproducibility.", "assessmentType": "evidence", "argument": "DeepSeek released V4.1-Flash-1017 weights under MIT and supplies local inference instructions, so users can download, inspect, and deploy the checkpoint themselves. Training data and the full training run remain less transparent, making this strong weight openness rather than complete reproducibility.", "score": 8, "href": "/models/deepseek-v4-1/#openness" }, "labour": { "summary": "DeepSeek’s V4 training-content summary says an internal team of professionals creates, labels, and audits data in some domains; the general training-method disclosure says its research team produces manually or automatically prepared optimization examples. Neither source identifies the exact labor behind V4.1-Flash-1017 or reports headcount, pay, employment terms, or working conditions.", "sourceUrls": [ "https://cdn.deepseek.com/policies/DeepSeek-V4-public-summary-of-training-content.pdf", "https://cdn.deepseek.com/policies/en-US/model-algorithm-disclosure.html" ], "confidence": "medium", "unknowns": "No worker counts, compensation, employment conditions, vendor information, or confirmation of labor practices specific to V4.1-Flash-1017.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "DeepSeek reports professional staff creating, labeling, and auditing some V4 data, confirming that human work contributes to the family. The sources do not identify the V4.1-Flash-1017 workforce or its terms, so the ethical quality of that labor remains unclear.", "score": null, "href": "/models/deepseek-v4-1/#labour" }, "military-contracts": { "summary": "DeepSeek publishes V4.1 Flash weights under MIT and documents local inference. That makes independent downstream deployments possible without a DeepSeek service contract; it is not evidence that a military customer has deployed them.", "sourceUrls": [ "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017" ], "confidence": "low", "unknowns": "Requires procurement and company source review.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "The exact checkpoint is MIT-licensed and its card gives local inference instructions. A downstream defense user could therefore run it without appearing in DeepSeek’s hosted-customer list; no source here identifies an actual military deployment, so that relationship remains unverified.", "score": null, "href": "/models/deepseek-v4-1/#military-contracts" }, "israel-palestine": { "summary": "We did not find DeepSeek among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "DeepSeek is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/deepseek-v4-1/#israel-palestine" } }, "sources": [ { "url": "https://api-docs.deepseek.com/updates/", "title": "DeepSeek API changelog", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://api-docs.deepseek.com/api/list-models/", "title": "Lists Models | DeepSeek API Docs", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://api-docs.deepseek.com/guides/search_fetch", "title": "Your First API Call | DeepSeek API Docs", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cdn.deepseek.com/policies/en-US/deepseek-privacy-policy.html", "title": "DeepSeek Privacy Policy (updated February 10, 2026)", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cdn.deepseek.com/policies/en-US/model-algorithm-disclosure.html", "title": "DeepSeek model mechanism and training methods disclosure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://api-docs.deepseek.com/api/create-response/", "title": "DeepSeek Responses API reference", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://api-docs.deepseek.com/news/news261017/", "title": "DeepSeek-V4.1-Flash-1017 release announcement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017", "title": "DeepSeek-V4.1-Flash-1017 model repository and card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cdn.deepseek.com/policies/DeepSeek-V4-public-summary-of-training-content.pdf", "title": "DeepSeek V4 public summary of training content", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/deepseek-v4-1/", "overall": { "score": 5.3, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 10 } }, { "id": "qwen-3-8", "name": "Qwen3.8 family", "category": "models", "organization": "Alibaba Cloud / Qwen team", "website": "https://qwen.ai/", "summary": "Qwen's official model organization lists Qwen3.8 and Qwen3.8-Flash-Next checkpoints, including a 2.4T-parameter family member; release status and terms should be checked per checkpoint.", "scope": "The release signal is the Qwen team's maintained Hugging Face organization. Other categories lack complete direct corporate disclosures in this pass and are explicitly gaps.", "releaseNote": "As checked October 23, 2026, the Qwen team organization lists Qwen3.8 checkpoints and Qwen3.8-Flash-Next updated in August 2026. The official release announcement and license for the flagship checkpoint were not verified here.", "dimensions": { "energy": { "summary": "Qwen3.8 model materials describe a 27B checkpoint, pre-training and post-training stages, architecture, and local serving, but give no total training electricity, water, or emissions measurement. Claimed architecture or latency efficiency is not a training-footprint result.", "sourceUrls": [ "https://huggingface.co/Qwen/Qwen3.8-Flash-Next", "https://github.com/QwenLM/Qwen3.8/blob/main/README.md" ], "confidence": "low", "unknowns": "Need model card and measured energy/compute disclosure.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The Qwen3.8 card describes training stages and gives a local serving recipe, but reports no energy or emissions boundary for training or a representative inference workload. A buyer can choose a local deployment and measure its own power use, while the provider’s training footprint remains undisclosed in these materials.", "score": null, "href": "/models/qwen-3-8/#energy" }, "training-rights": { "summary": "Qwen’s official training-data summary names public internet material, non-public partner data, data-labeling services, paid contractors, and synthetic data; it says corpora may include copyrighted material and users can request training opt-out. This is meaningful category-level disclosure, but it does not list datasets, licenses, rightsholders, or the opt-out’s scope and processing.", "sourceUrls": [ "https://qwen.ai/training-data-summary", "https://huggingface.co/Qwen" ], "confidence": "medium", "unknowns": "No source-level Qwen3.8 corpus manifest, license ledger, independent rights audit, or detailed opt-out procedure was found.", "suggestedScore": 4, "scoreReason": "Qwen discloses source categories and an opt-out route, but also says training may include copyrighted data without giving itemized permissions.", "assessmentType": "conditional", "argument": "Qwen’s official summary identifies public, partner, contractor, and synthetic data, and acknowledges that training may include copyrighted material. It does not tie particular sources or licenses to Qwen3.8; request checkpoint-specific provenance and details of the opt-out route.", "score": 4, "href": "/models/qwen-3-8/#training-rights" }, "privacy": { "summary": "Qwen Studio terms (updated May 19, 2026) define prompts and outputs as user content, treat them as non-confidential absent a separate agreement, and permit use of non-personal user content to develop AI technologies; the training-data summary says users may request opt-out. These terms govern Qwen Studio, not locally hosted weights or every third-party/API deployment.", "sourceUrls": [ "https://qwen.ai/termsservice" ], "confidence": "high", "unknowns": "The current Studio privacy policy and requested opt-out execution/retention details must be checked for the account and jurisdiction; these terms do not establish controls for self-hosted Qwen.", "suggestedScore": 3, "scoreReason": "First-party terms expressly treat prompts as non-confidential and permit some user content to be used for model development, with a separate opt-out route disclosed.", "assessmentType": "evidence", "argument": "The Studio terms expressly allow AI-development use of non-personal user content and disclaim confidentiality absent a separate agreement, while a separate summary offers an opt-out request. For Qwen Studio, avoid confidential prompts unless a separate agreement provides stronger protection; this finding does not apply to self-hosted weights.", "score": 3, "href": "/models/qwen-3-8/#privacy" }, "openness": { "summary": "Qwen checkpoints are published through the Qwen team organization, but model availability alone does not establish license scope or full training-data openness. Verify the exact checkpoint's license before use.", "sourceUrls": [ "https://huggingface.co/Qwen" ], "confidence": "medium", "unknowns": "The listed collection does not establish the exact license or full recipe for each Qwen3.8 checkpoint.", "suggestedScore": 5, "scoreReason": "Official team checkpoint listings are public, but exact checkpoint license and training-data openness remain unverified.", "assessmentType": "conditional", "argument": "The Qwen organization publishes checkpoints, which enables download and inspection, but the listing alone does not establish every checkpoint’s license or release of training data. Verify the exact model card and license before reuse; weights availability is only one part of openness.", "score": 5, "href": "/models/qwen-3-8/#openness" }, "labour": { "summary": "Qwen’s training-data summary expressly identifies data-labeling services and paid contractors, and describes human review and annotation frameworks. It does not disclose worker numbers, locations, compensation, employment status, labor protections, or vendor names.", "sourceUrls": [ "https://qwen.ai/training-data-summary" ], "confidence": "medium", "unknowns": "No workforce or contractor terms are disclosed in the source; no independent investigation of Qwen labeling vendors was located.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "evidence", "argument": "Qwen confirms that paid contractors and data-labeling services contribute to its training data, so labor is not merely hypothetical. The summary says nothing about those workers’ pay or conditions, which leaves worker conditions unknown; ask for vendor standards and audit results.", "score": null, "href": "/models/qwen-3-8/#labour" }, "military-contracts": { "summary": "Qwen3.8-Flash-Next weights are publicly downloadable under Qwen Community License 1.0 with local inference instructions. Independent downstream use is possible under that license; the checked release sources do not document a military deployment.", "sourceUrls": [ "https://huggingface.co/Qwen/Qwen3.8-Flash-Next" ], "confidence": "low", "unknowns": "Needs procurement and corporate announcement review.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "The exact Qwen3.8-Flash-Next card provides downloadable weights and local serving instructions under the Qwen Community License 1.0. This permits downstream deployment subject to that license, so vendor contracting alone cannot reveal every use; these sources do not establish actual military use.", "score": null, "href": "/models/qwen-3-8/#military-contracts" }, "israel-palestine": { "summary": "We did not find Alibaba Cloud / Qwen team among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Alibaba Cloud / Qwen team is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/qwen-3-8/#israel-palestine" } }, "sources": [ { "url": "https://huggingface.co/Qwen", "title": "Qwen team models and releases", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://qwenlm.github.io/blog/qwen3/", "title": "Qwen3: Think Deeper, Act Faster", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://qwen.ai/termsservice", "title": "Qwen Studio Terms of Service (last updated May 19, 2026)", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://qwen.ai/training-data-summary", "title": "Qwen training data summary", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next", "title": "Qwen3.8-Flash-Next model card and local deployment instructions", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/QwenLM/Qwen3.8/blob/main/README.md", "title": "Qwen3.8 official releases", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/qwen-3-8/", "overall": { "score": 4, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 14 } }, { "id": "mistral-large-4", "name": "Mistral Large 4", "category": "models", "organization": "Mistral AI", "website": "https://mistral.ai/", "summary": "Frontier 1T-parameter MoE model in public API preview; Mistral said weights were expected at the end of October 2026, so treat it as hosted/API at this check date.", "scope": "Release facts refer to Large 4. Company sustainability and employment policies were not verified; do not project Mistral Small 4 openness onto Large 4 before weights and license are actually published.", "releaseNote": "Mistral announced public preview of Large 4 on October 6, 2026 and said weights would drop at month end. As of October 23, the announcement supports preview availability, not verified weight publication.", "dimensions": { "energy": { "summary": "Disclosure gap: Large 4 announcement gives size and model capabilities but no training or inference energy, water, or emissions. Sparse active-parameter counts are not an energy audit.", "sourceUrls": [ "https://mistral.ai/news/mistral-large-4/" ], "confidence": "low", "unknowns": "No model-specific energy measurement located.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The Large 4 announcement gives architecture and size information, not training energy, emissions, or inference measurements. Sparse active parameters do not prove low total energy; request measured training and serving data before comparing footprint.", "score": null, "href": "/models/mistral-large-4/#energy" }, "training-rights": { "summary": "Disclosure gap: the release page does not provide an itemized training corpus or rights/consent ledger for Large 4.", "sourceUrls": [ "https://mistral.ai/news/mistral-large-4/" ], "confidence": "low", "unknowns": "Need full model card, datasets, license and opt-out/rights policy.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Mistral’s release page does not identify the Large 4 training corpus or explain its rights process. Without those details, users cannot verify the source permissions; ask Mistral for dataset provenance.", "score": null, "href": "/models/mistral-large-4/#training-rights" }, "privacy": { "summary": "Mistral's privacy policy (effective Sep. 3, 2026) covers consumer Mistral AI Products, including Mistral AI Studio, and expressly excludes business-activity processing where Mistral acts as processor. For covered consumer products, Inputs/Outputs may be used for model training subject to an account opt-out; API input/output is generally retained through generation plus 30 rolling days for abuse monitoring unless zero-data-retention is activated, while Agents API content is retained until account termination. This is company-product policy, not a Large 4 model-card property; commercial/API deployment terms may differ.", "sourceUrls": [ "https://legal.mistral.ai/terms/privacy-policy/" ], "confidence": "high", "unknowns": "Confirm service-specific terms and configured ZDR/retention for any particular Large 4 API deployment; policy excludes business processing and does not govern self-hosted weights.", "suggestedScore": 4, "scoreReason": "The consumer policy provides an account-level training opt-out and optional zero-data-retention setting, but otherwise permits training on covered Inputs/Outputs and a default 30-day API abuse-monitoring retention; score is limited to those terms.", "assessmentType": "conditional", "argument": "Mistral’s current privacy policy covers consumer products and explicitly separates business processing, so the consumer terms cannot be carried over to Studio/API use. Evaluate the exact product contract and retention controls before using confidential prompts.", "score": 4, "href": "/models/mistral-large-4/#privacy" }, "openness": { "summary": "Large 4 is in public preview API; the release says weights are forthcoming at end of October. It is not verified as open-weight on the check date. Mistral's Apache-licensed Small 4 is a separate release and cannot establish Large 4 terms.", "sourceUrls": [ "https://mistral.ai/news/mistral-large-4/", "https://mistral.ai/news/mistral-small-4/" ], "confidence": "high", "unknowns": "Check whether weights were actually released after October 23 and inspect the exact Large 4 license.", "suggestedScore": 2, "scoreReason": "As of the checked release announcement, Large 4 was API preview with weights promised later, not yet verified.", "assessmentType": "conditional", "argument": "At the check date, Large 4 was in API preview and the announcement said weights would follow later in October; the record does not verify that release. At the check date, treat the checkpoint as hosted; revisit this only after confirming the weights and license are available.", "score": 2, "href": "/models/mistral-large-4/#openness" }, "labour": { "summary": "Disclosure gap: the checked Large 4 announcement does not describe human data workers, annotator pay, contractor conditions, or labour protections.", "sourceUrls": [ "https://mistral.ai/news/mistral-large-4/" ], "confidence": "low", "unknowns": "No model-specific labour disclosure found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The Large 4 announcement gives no information on data workers or contractor conditions. The announcement leaves worker treatment unknown; request workforce and vendor disclosures.", "score": null, "href": "/models/mistral-large-4/#labour" }, "military-contracts": { "summary": "Mistral announced a November 2025 partnership with Helsing to develop vision-language-action models for defense and security applications. This documents a company/model-family defense partnership, not a Mistral Large 4 deployment or contract.", "sourceUrls": [ "https://mistral.ai/news/ki-fur-deutschland/" ], "confidence": "high", "unknowns": "The announcement concerns development of specialized VLA systems and does not name Large 4 or disclose a complete defense customer/deployment register.", "suggestedScore": 4, "scoreReason": "A defense partnership is disclosed at company level, but the evidence does not attribute military deployment to the Large 4 general-purpose checkpoint.", "assessmentType": "conditional", "argument": "Mistral’s Helsing announcement documents a defense/security partnership to develop specialized VLA systems, not a Large 4 deployment. It is a company-level concern, but buyers should not attribute that work to the general-purpose checkpoint without a version-specific link.", "score": 4, "href": "/models/mistral-large-4/#military-contracts" }, "israel-palestine": { "summary": "We did not find Mistral AI among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Mistral AI is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/mistral-large-4/#israel-palestine" } }, "sources": [ { "url": "https://mistral.ai/news/mistral-large-4/", "title": "Introducing Mistral Large 4 | Mistral", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://mistral.ai/news/mistral-small-4/", "title": "Introducing Mistral Small 4", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://mistral.ai/news/", "title": "Mistral latest news", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://legal.mistral.ai/terms/privacy-policy/", "title": "Mistral AI Privacy Policy (effective September 3, 2026)", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://mistral.ai/news/ki-fur-deutschland/", "title": "KI für Deutschland", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/mistral-large-4/", "overall": { "score": 3.3, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "red", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 20 } }, { "id": "phi-4", "name": "Phi-4 family", "category": "models", "organization": "Microsoft", "website": "https://azure.microsoft.com/products/phi/", "summary": "Compact open model family; Microsoft's official product page calls Phi-4-mini and Phi-4-multimodal its newest Phi models. No newer flagship release was verified in this pass.", "scope": "Product listing establishes the latest verified family members as of this check, not that no later checkpoint exists. Environmental/privacy/customer findings are not model-specific unless noted.", "releaseNote": "Microsoft's Phi product page identifies Phi-4-mini and Phi-4-multimodal as the newest members. Checked October 23, 2026; no newer Phi release was verified through the official page.", "dimensions": { "energy": { "summary": "Microsoft positions Phi as small language models, which enables local deployment options, but the checked official page provides no Phi-4 training or inference energy/water/carbon measurement.", "sourceUrls": [ "https://azure.microsoft.com/en-us/products/phi" ], "confidence": "low", "unknowns": "No model-specific lifecycle or standardized task-energy data found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "Microsoft describes Phi as small models suited to local deployment but provides no Phi-4 energy measurement in the cited materials. Local use can change where energy is consumed rather than prove that less is used; compare the target device and workload.", "score": null, "href": "/models/phi-4/#energy" }, "training-rights": { "summary": "Disclosure gap: the product page describes capabilities and model family but not a Phi-4-specific training corpus, licensing, or consent audit.", "sourceUrls": [ "https://azure.microsoft.com/en-us/products/phi", "https://huggingface.co/collections/microsoft/phi-4", "https://github.com/microsoft/PhiCookBook/blob/main/md/01.Introduction/01/01.PhiFamily.md" ], "confidence": "low", "unknowns": "Need model cards and data provenance for exact Phi variants.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Microsoft’s Phi product pages, collection, and cookbook explain the family but do not provide a Phi-4 training-corpus list or rights audit. Users who need source traceability should request checkpoint-level documentation.", "score": null, "href": "/models/phi-4/#training-rights" }, "privacy": { "summary": "For Phi models that are deployed as 'Models sold by Azure' in Microsoft Foundry, Microsoft's current policy says prompts/completions are not used to train or improve foundation models, and are not stored in the stateless model; standard deployments process data in the customer-specified geography. Abuse monitoring can retain flagged prompt/completion samples for limited human review unless modified monitoring is approved. Stateful features (such as Responses/Threads/files) may store data in the Azure tenant. Microsoft's catalog lists Phi-4-family models in its Azure-sold availability material. These protections do not automatically apply to partner/community listings, other hosts, or local deployments.", "sourceUrls": [ "https://learn.microsoft.com/en-us/azure/foundry/responsible-ai/openai/data-privacy", "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure-region-availability?pivots=standard" ], "confidence": "high", "unknowns": "The score applies only to a Phi deployment explicitly designated sold by Azure and configured deployment type/features; partner/community hosted models, local installations, and preview exceptions can differ.", "suggestedScore": 7, "scoreReason": "Azure's sold-model policy provides strong no-foundation-model-training and geography controls for standard deployments, while abuse monitoring can include human review and stateful features retain data; the rating is conditional on the exact Azure deployment class.", "assessmentType": "conditional", "argument": "Azure Foundry’s policy says prompts and completions for models sold directly by Azure are not used to train or improve foundation models, subject to the listed service boundary. This protection does not automatically cover other hosts or Microsoft products, so verify the exact route and region.", "score": 7, "href": "/models/phi-4/#privacy" }, "openness": { "summary": "Microsoft labels Phi as an open model family and provides deployment documentation; access terms and weights are checkpoint-specific. Verify the license attached to the chosen Phi-4 variant.", "sourceUrls": [ "https://azure.microsoft.com/en-us/products/phi", "https://huggingface.co/collections/microsoft/phi-4", "https://github.com/microsoft/PhiCookBook/blob/main/md/01.Introduction/01/01.PhiFamily.md" ], "confidence": "medium", "unknowns": "No exact Phi-4 license or complete training recipe was checked here.", "suggestedScore": 6, "scoreReason": "Microsoft presents Phi as an open model family, but this pass did not verify the exact Phi-4 checkpoint license and artifacts.", "assessmentType": "conditional", "argument": "Microsoft calls Phi an open model family and publishes some weights and deployment materials, but the exact checkpoint license and artifacts vary. This reflects access to selected weights, not a claim that every Phi release or its training dataset is open.", "score": 6, "href": "/models/phi-4/#openness" }, "labour": { "summary": "Company context: Microsoft. Microsoft’s FY2025 Human Rights Transparency Report describes supplier monitoring and reports findings in its Devices/Cloud Supply Chain and Provisioning scope, including 34 forced-labor/trafficking indicators (33 closed, one corrective plan), 50 fair-compensation findings (46 closed, four plans), and 75 working-hours/rest findings (68 closed, seven plans). These are supplier-audit findings, not AI-annotation-specific results. This is a record of the company behind Phi-4 family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Human-Rights-Transparency-Report-2025.pdf", "https://huggingface.co/microsoft/Phi-4" ], "confidence": "high", "unknowns": "Working conditions for the people building or supporting Phi-4 family are not established by this company-wide record. The published findings cover specified supplier audit scopes, not all suppliers, model evaluation labor, or all remediation outcomes.", "suggestedScore": 6, "scoreReason": "Company-level assessment: Microsoft reports supplier findings and closure plans, but the reported scope is not specific to AI annotation and some plans remain open. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "Microsoft reports supplier findings and closure plans, but the reported scope is not specific to AI annotation and some plans remain open. This is relevant when deciding whose business to support, while pay and working conditions for the team behind Phi-4 family remain unestablished. Ask for current worker outcomes for that team and its suppliers.", "assessmentScope": "Company record", "productArgument": "The Phi-4 card documents model provenance and licensing, not the conditions of people who prepared data or evaluations. Its technical openness cannot substitute for labor terms; users comparing worker treatment should treat this category as undocumented for Phi-4.", "score": 6, "href": "/models/phi-4/#labour" }, "military-contracts": { "summary": "Company context: Microsoft. Microsoft’s final 2026 review says its external investigation found evidence supporting elements of 2025 reporting, including IMOD Azure storage consumption in the Netherlands and AI-service use. Microsoft says it disabled specified storage/AI subscriptions in September 2025, while continuing other software, professional services, Azure and Azure AI services to IMOD. This is a record of the company behind Phi-4 family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Summary-of-2025-External-Investigation-Follow-Up.pdf", "https://huggingface.co/microsoft/Phi-4", "https://www.microsoft.com/en-us/us-government/defense-intelligence" ], "confidence": "high", "unknowns": "Deployment of Phi-4 family under these contracts is not established. The report does not access IMOD customer content, does not establish the content/use of every service, and does not say the wider commercial relationship ended.", "suggestedScore": 2, "scoreReason": "Company-level assessment: The company confirms defense-ministry Azure/AI use and a review supporting elements of serious surveillance-related reporting; it stopped specified services but continues others. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "The company confirms defense-ministry Azure/AI use and a review supporting elements of serious surveillance-related reporting; it stopped specified services but continues others. This is relevant when deciding whose business to support. Deployment of Phi-4 family or this particular model release under those contracts remains unestablished; consider the company relationship separately from the product’s actual use.", "assessmentScope": "Company record", "productArgument": "Microsoft markets mission-ready cloud and AI for U.S. defense and intelligence, while its Phi-4 materials establish only the model release and license. That makes the parent company’s defense business relevant context, but a deployment record must name Phi-4 before attributing military use to this checkpoint.", "score": 2, "href": "/models/phi-4/#military-contracts" }, "israel-palestine": { "summary": "Microsoft is named as a consumer boycott priority target in the BNC guide. The product is developed or supplied by Microsoft.", "argument": "If following this BNC campaign is a requirement, consider alternatives to paid Microsoft products where practical. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well. This label follows the named company’s campaign status; it does not claim that BDS separately reviewed this exact product.", "assessmentType": "evidence", "confidence": "high", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://bdsmovement.net/microsoft" ], "bdsStatus": "priority-boycott", "statusLabel": "Boycott target", "reviewedAt": "2026-10-23", "assessmentScope": "Developer / service company", "bdsMatch": { "target": "Microsoft", "status": "priority-boycott", "scope": "Developer / service company", "relationship": "The product is developed or supplied by Microsoft.", "application": "If following this BNC campaign is a requirement, consider alternatives to paid Microsoft products where practical.", "sourceIds": [ "microsoft" ] }, "upstreamExposure": null, "businessContext": { "summary": "Company context: Microsoft. Microsoft’s final 2026 review says Covington found evidence supporting elements of reporting about IMOD Azure storage and AI-service use, including Azure storage consumption in the Netherlands. Microsoft says it disabled specified subscriptions/services in September 2025 but continues providing IMOD with other Azure, Azure AI, software, and professional services. The report did not access customer content and does not establish all downstream uses. This is a record of the company behind Phi-4 family; it does not establish use of this particular product or release.", "unknowns": "How Phi-4 family is used in this relationship is not established. Microsoft describes only selected service suspensions; the report does not establish all IMOD use, civilian impact, or the full scope of continuing services.", "sourceUrls": [ "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Summary-of-2025-External-Investigation-Follow-Up.pdf" ] }, "score": null, "href": "/models/phi-4/#israel-palestine" } }, "sources": [ { "url": "https://azure.microsoft.com/en-us/products/phi", "title": "Phi open models", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/collections/microsoft/phi-4", "title": "Phi-4 | Microsoft model collection", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/microsoft/PhiCookBook/blob/main/md/01.Introduction/01/01.PhiFamily.md", "title": "Microsoft's Phi family | Phi Cookbook", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://learn.microsoft.com/en-us/azure/foundry/responsible-ai/openai/data-privacy", "title": "Microsoft Foundry data, privacy, and security for Models sold by Azure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure-region-availability?pivots=standard", "title": "Microsoft Foundry region availability for Models sold by Azure (Phi-4 family)", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/microsoft/Phi-4", "title": "Microsoft Phi-4 model card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.microsoft.com/en-us/us-government/defense-intelligence", "title": "Microsoft U.S. defense and intelligence offerings", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Human-Rights-Transparency-Report-2025.pdf", "title": "Microsoft Human Rights Transparency Report 2025", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Summary-of-2025-External-Investigation-Follow-Up.pdf", "title": "Microsoft Summary of 2025 External Investigation and Follow Up", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/microsoft", "title": "BNC: Microsoft campaign and subsidiaries guidance", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/phi-4/", "overall": { "score": 5.3, "rated": 4, "total": 6, "coverage": 0.6666666666666666, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 9 } }, { "id": "command-a-plus", "name": "Command A+", "category": "models", "organization": "Cohere", "website": "https://cohere.com/command", "summary": "218B-parameter, 25B-active MoE enterprise model released under Apache 2.0, deployable on-premises or in customer-controlled environments.", "scope": "Model specifications/license are release-specific. Vendor's enterprise privacy/security posture does not establish every customer's deployment configuration.", "releaseNote": "Cohere announced Command A+ on May 20, 2026 and lists model ID command-a-plus-05-2026 under Apache 2.0. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "Cohere says Command A+ can run on one B200 or two H100s and promotes efficiency, but publishes no energy, water, or emissions per training run or task. Hardware minimum is not a measured footprint.", "sourceUrls": [ "https://cohere.com/blog/command-a-plus", "https://docs.cohere.com/v2/changelog" ], "confidence": "medium", "unknowns": "No model-specific energy measurement or independent validation found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "Cohere publishes the hardware footprint needed for deployment and promotes efficiency, but not measured model energy or emissions. Hardware requirements bound a possible deployment; actual footprint still depends on utilization, electricity mix, and workload.", "score": null, "href": "/models/command-a-plus/#energy" }, "training-rights": { "summary": "The release and changelog provide technical specifications and license but no itemized training corpus or rights/consent register for Command A+.", "sourceUrls": [ "https://cohere.com/blog/command-a-plus", "https://docs.cohere.com/v2/changelog" ], "confidence": "low", "unknowns": "Training-data provenance, licenses, and opt-out handling remain undisclosed in sources checked.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The Command A+ release and changelog describe model features and licensing but not the training corpus or consent process. Those technical details do not establish source permissions; request provenance and opt-out information.", "score": null, "href": "/models/command-a-plus/#training-rights" }, "privacy": { "summary": "Cohere promotes private deployment and says its models can run in a customer's VPC, on-premises, or air-gapped environment. Actual privacy depends on selected hosting and telemetry configuration; this is not a blanket API retention guarantee.", "sourceUrls": [ "https://cohere.com/blog/cohere-releases-command-a-plus" ], "confidence": "medium", "unknowns": "Need service-specific retention and training-use terms for API and private deployment.", "suggestedScore": 7, "scoreReason": "The release supports private, on-premises and air-gapped deployment, subject to operator configuration.", "assessmentType": "conditional", "argument": "Cohere supports VPC, on-premises, and air-gapped deployment, which can reduce exposure to an external host when configured by the customer. Those options do not prove isolation by themselves; assess logging, access, and retention in the selected deployment.", "score": 7, "href": "/models/command-a-plus/#privacy" }, "openness": { "summary": "Cohere says Command A+ is available under Apache 2.0 and offers model weights; its complete pretraining data and training process are not shown as fully reproducible in the release page.", "sourceUrls": [ "https://cohere.com/blog/command-a-plus", "https://docs.cohere.com/v2/changelog" ], "confidence": "high", "unknowns": "Open weights/license do not establish open training data or reproducibility.", "suggestedScore": 7, "scoreReason": "Apache 2.0 weights and self-deployment are documented; the training corpus and full reproducible pipeline are not disclosed.", "assessmentType": "evidence", "argument": "Cohere offers weights under Apache 2.0, giving broad reuse and local deployment rights. Because the training data and full recipe remain undisclosed, this is open-weight rather than fully reproducible openness.", "score": 7, "href": "/models/command-a-plus/#openness" }, "labour": { "summary": "Cohere says Command A+ training used proprietary datasets with human annotation, automated and synthetic data, plus specialist data vendors. It does not disclose annotator pay, working conditions, or protections.", "sourceUrls": [ "https://docs.cohere.com/docs/command-a-plus" ], "confidence": "low", "unknowns": "Annotator compensation, working conditions, exposure controls, vendor audits and current outcomes remain undisclosed in the reviewed documentation.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Cohere’s exact Command A+ documentation confirms both human annotation and vendor-sourced datasets, so labor is part of this model’s pipeline. Because that disclosure omits compensation and protections, it supports asking about supplier standards but not a conclusion about worker treatment.", "score": null, "href": "/models/command-a-plus/#labour" }, "military-contracts": { "summary": "Cohere says its MOU with defense-focused Indra includes Cohere sovereign LLMs for IndraMind and a defense cooperation line for mission-planning and multinational operations. The announcement does not name Command A+ as the deployed model or identify a completed contract.", "sourceUrls": [ "https://cohere.com/blog/cohere-announces-strategic-mous-with-indragroup-and-multiverse-computing", "https://docs.cohere.com/docs/command-a-plus" ], "confidence": "medium", "unknowns": "The MOU does not establish a completed procurement, Command A+ deployment, operational use or independent oversight of the proposed defense work.", "suggestedScore": null, "scoreReason": "The company describes proposed defense cooperation under an MOU. That supports a relationship finding, but the reviewed evidence does not establish a completed deployment or enough about enforceable safeguards for a numerical judgment.", "assessmentType": "conditional", "argument": "Cohere’s May 2026 announcement links its LLMs to IndraMind and describes defense cooperation around analysis, mission planning, and interoperability. This is a documented company-level defense relationship, but not proof that Command A+ itself is in use; the model-specific link remains unestablished.", "assessmentScope": "Company record", "score": null, "href": "/models/command-a-plus/#military-contracts" }, "israel-palestine": { "summary": "We did not find Cohere among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Cohere is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/command-a-plus/#israel-palestine" } }, "sources": [ { "url": "https://cohere.com/blog/command-a-plus", "title": "Introducing Command A+", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.cohere.com/v2/changelog", "title": "Cohere release notes", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cohere.com/blog/cohere-releases-command-a-plus", "title": "Command A+ for sovereign AI", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.cohere.com/v1/docs/models", "title": "An overview of Cohere models", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.cohere.com/docs/command-a-plus", "title": "Cohere Command A+ model data and training documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cohere.com/blog/cohere-announces-strategic-mous-with-indragroup-and-multiverse-computing", "title": "Cohere strategic MOU with Indra Group including defense cooperation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/command-a-plus/", "overall": { "score": 7, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 6 } }, { "id": "amazon-nova-2", "name": "Amazon Nova 2 family", "category": "models", "organization": "Amazon / AWS", "website": "https://aws.amazon.com/nova/", "summary": "Hosted foundation-model portfolio including Nova 2 Lite, Nova 2.5 Sonic, and multimodal embeddings. The flagship general-purpose text/reasoning item surfaced in official docs is Nova 2 Lite.", "scope": "Model offerings are AWS-specific. Energy and Project Nimbus findings are Amazon/AWS-level and do not establish Nova use in an Israeli government workload.", "releaseNote": "AWS documentation and its October 22, 2026 update identify Nova 2 Lite and Nova 2.5 Sonic among current offerings. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "Amazon reports AWS-wide data-center efficiency and energy matching, but provides no Nova-specific training or inference energy, water, or carbon accounting. Corporate portfolio statistics cannot be attributed to a single Nova request.", "sourceUrls": [ "https://sustainability.aboutamazon.com/products-services/aws-cloud?energy=true", "https://www.aboutamazon.com/impact/amazon-data-center-tenets" ], "confidence": "medium", "unknowns": "No Nova-specific compute or lifecycle measurements.", "suggestedScore": null, "scoreReason": "No score: the available sources report hardware, throughput, or provider-wide efficiency framing, but no model-specific energy, water, or emissions impact. Efficiency claims alone do not establish a lower footprint.", "assessmentType": "conditional", "argument": "Amazon publishes AWS data-center efficiency and electricity-matching information, but those corporate figures do not measure Nova 2 training or a Nova request. These disclosures provide infrastructure context, but do not establish Nova-specific lifecycle impact.", "score": null, "href": "/models/amazon-nova-2/#energy" }, "training-rights": { "summary": "Disclosure gap: Nova product docs describe capabilities but do not provide model-specific training dataset sources, rights, or licensing detail.", "sourceUrls": [ "https://docs.aws.amazon.com/nova/latest/nova2-userguide/whats-new.html" ], "confidence": "low", "unknowns": "Need a Nova model card/data provenance report and rights audit.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Nova 2 product documentation describes service capabilities but gives no training-data sources or rights account. Users cannot verify the corpus permissions from these materials and should request a model-specific provenance statement.", "score": null, "href": "/models/amazon-nova-2/#training-rights" }, "privacy": { "summary": "Nova is served through AWS Bedrock and related AWS services, so customer data handling depends on AWS service terms and configuration. The product page alone does not fully establish Nova-specific retention or training policy.", "sourceUrls": [ "https://aws.amazon.com/nova/models/", "https://docs.aws.amazon.com/nova/latest/nova2-userguide/whats-new.html" ], "confidence": "medium", "unknowns": "Need current Bedrock data-use, logging, retention, region, and human review terms for the selected endpoint.", "suggestedScore": 5, "scoreReason": "Managed Bedrock controls exist, but exact endpoint-specific retention and training terms were not verified here.", "assessmentType": "conditional", "argument": "Nova runs through Bedrock and related AWS services, so data handling depends on the selected service, region, retention configuration, and AWS terms. Do not treat a product catalog as a privacy guarantee; inspect the applicable Bedrock data controls.", "score": 5, "href": "/models/amazon-nova-2/#privacy" }, "openness": { "summary": "Nova 2 is distributed as a managed service through Amazon Bedrock; no public model weights or full training recipe were identified for Nova 2 Lite in the checked docs.", "sourceUrls": [ "https://aws.amazon.com/nova/models/" ], "confidence": "high", "unknowns": "No public weights or full training artifacts found.", "suggestedScore": 2, "scoreReason": "Managed service access was verified without public weights or a reproducible training stack.", "assessmentType": "evidence", "argument": "Nova 2 is offered as a managed Bedrock service; the checked documentation does not provide downloadable weights or a reproducible training package. Customers can use the service but cannot independently inspect or host the checkpoint.", "score": 2, "href": "/models/amazon-nova-2/#openness" }, "labour": { "summary": "Company context: Amazon / AWS. Amazon’s Supply Chain Standards Manual makes human-rights and labor standards a condition of supplier relationships and describes monitoring and corrective processes. This is a supply-chain policy, not evidence of current Bedrock annotation-worker conditions or audit outcomes. This is a record of the company behind Amazon Nova 2 family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://sustainability.aboutamazon.com/amazon_supplier_manual_english.pdf", "https://aws.amazon.com/nova/models/" ], "confidence": "high", "unknowns": "Working conditions for the people building or supporting Amazon Nova 2 family are not established by this company-wide record. No current Bedrock- or foundation-model-specific labor audit results were established.", "suggestedScore": 4, "scoreReason": "Company-level assessment: Amazon requires supplier standards and describes monitoring, but publishes no Bedrock-specific worker outcomes here. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "Amazon requires supplier standards and describes monitoring, but publishes no Bedrock-specific worker outcomes here. This is relevant when deciding whose business to support, while pay and working conditions for the team behind Amazon Nova 2 family remain unestablished. Ask for current worker outcomes for that team and its suppliers.", "assessmentScope": "Company record", "productArgument": "The Nova catalog explains model features and customer customization, not the workforce behind foundation-model training or evaluation. That leaves worker conditions unassessed; customer-side fine-tuning should not be mistaken for evidence about Amazon’s training suppliers.", "score": 4, "href": "/models/amazon-nova-2/#labour" }, "military-contracts": { "summary": "Company context: Amazon / AWS. AWS funds U.S. defense work that names autonomous systems and battle-management uses. Separately, Israel’s procurement administration identifies AWS as a Nimbus public-cloud provider; AFSC and Who Profits describe military/security use, while the Guardian reports contract terms limiting service withdrawal. These sources do not establish AWS’s role in a particular attack. This is a record of the company behind Amazon Nova 2 family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://aws.amazon.com/blogs/publicsector/aws-announces-up-to-100-million-in-federal-credits-to-accelerate-innovation-for-national-security-and-scientific-missions/", "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/amazon", "https://aws.amazon.com/nova/models/", "https://aws.amazon.com/federal/defense/" ], "confidence": "high", "unknowns": "Deployment of Amazon Nova 2 family under these contracts is not established. The announced credits and eligible use cases do not establish which projects received funding or which AWS AI services each project selected.", "suggestedScore": 2, "scoreReason": "Company-level assessment: AWS funds defense work that explicitly includes autonomous systems and battle management across classification levels; no independent safeguards assessment is published. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "conditional", "argument": "AWS funds defense work that explicitly includes autonomous systems and battle management across classification levels; no independent safeguards assessment is published. This is relevant when deciding whose business to support. Deployment of Amazon Nova 2 family or this particular model release under those contracts remains unestablished; consider the company relationship separately from the product’s actual use.", "assessmentScope": "Company record", "productArgument": "AWS states that its infrastructure supports DoD missions and names DoD-specific cloud offerings. Nova 2 is separately described as a Bedrock model, but no cited source ties that model to a military customer; treat AWS defense business as context, not model deployment evidence.", "score": 2, "href": "/models/amazon-nova-2/#military-contracts" }, "israel-palestine": { "summary": "Amazon is named as a pressure target in the BNC’s official campaign sources. The product is developed or supplied by Amazon.", "argument": "If following this BNC campaign is a requirement, consider alternatives to paid Amazon products where practical. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well. This label follows the named company’s campaign status; it does not claim that BDS separately reviewed this exact product.", "assessmentType": "evidence", "confidence": "high", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "pressure", "statusLabel": "Pressure target", "reviewedAt": "2026-10-23", "assessmentScope": "Developer / service company", "bdsMatch": { "target": "Amazon", "status": "pressure", "scope": "Developer / service company", "relationship": "The product is developed or supplied by Amazon.", "application": "If following this BNC campaign is a requirement, consider alternatives to paid Amazon products where practical.", "sourceIds": [] }, "upstreamExposure": null, "businessContext": { "summary": "Company context: Amazon / AWS. Israel’s procurement administration names AWS as a Nimbus public-cloud provider and the Ministry of Defense and IDF as project partners. The Guardian’s investigation of leaked agreement documents reports terms restricting AWS’s ability to suspend service, including after alleged policy violations. Who Profits and AFSC describe AWS cloud services in military/security infrastructure and surveillance-related uses; those are attributed research claims, not proof of AWS’s role in a specific attack. Both companies denied circumventing legal obligations in response to the Guardian investigation. This is a record of the company behind Amazon Nova 2 family; it does not establish use of this particular product or release.", "unknowns": "How Amazon Nova 2 family is used in this relationship is not established. The government record establishes the contract and project participants; the investigative reporting describes contract controls, and research organizations make broader use claims. The reviewed evidence does not establish AWS’s contribution to a particular operation, injury, or other specific harm.", "sourceUrls": [ "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/amazon", "https://aws.amazon.com/blogs/aws/now-open-aws-israel-tel-aviv-region/", "https://aws.amazon.com/local/israel/" ] }, "score": null, "href": "/models/amazon-nova-2/#israel-palestine" } }, "sources": [ { "url": "https://aws.amazon.com/nova/models/", "title": "Amazon Nova models", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.aws.amazon.com/nova/latest/nova2-userguide/whats-new.html", "title": "What's new in Amazon Nova 2", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.aboutamazon.com/news/aws/amazon-nova-foundation-models-guide", "title": "Everything you need to know about Amazon Nova", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://sustainability.aboutamazon.com/products-services/aws-cloud?energy=true", "title": "AWS Cloud sustainability", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.aboutamazon.com/impact/amazon-data-center-tenets", "title": "Amazon's approach to data centers", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://aws.amazon.com/blogs/aws/now-open-aws-israel-tel-aviv-region/", "title": "Now open: AWS Israel (Tel Aviv) Region", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://aws.amazon.com/local/israel/", "title": "AWS Israel region", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://aws.amazon.com/federal/defense/", "title": "AWS U.S. defense cloud offerings", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://sustainability.aboutamazon.com/amazon_supplier_manual_english.pdf", "title": "Amazon Supply Chain Standards Manual", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://aws.amazon.com/blogs/publicsector/aws-announces-up-to-100-million-in-federal-credits-to-accelerate-innovation-for-national-security-and-scientific-missions/", "title": "AWS announces up to $100 million in federal credits for national security and scientific missions", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://mr.gov.il/ilgstorefront/en/news/details/111222", "title": "Nimbus Tender — Israel Government Procurement Administration", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "title": "Revealed: Israel demanded Google and Amazon use secret “wink” to sidestep legal orders", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://www.whoprofits.org/publications/report/164", "title": "Digitizing Occupation: The Role of Big Tech in Israeli Military Infrastructure", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://investigate.afsc.org/company/amazon", "title": "Amazon — AFSC Investigate", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/amazon-nova-2/", "overall": { "score": 3.3, "rated": 4, "total": 6, "coverage": 0.6666666666666666, "partial": true, "color": "red", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 18 } }, { "id": "nemotron-3-5", "name": "Nemotron 3.5 family", "category": "models", "organization": "NVIDIA", "website": "https://www.nvidia.com/en-us/ai-data-science/foundation-models/", "summary": "Open model family for agent workflows. Latest verified general-purpose release in this pass is Nemotron 3.5 Super VL (120B) alongside Bolt edge variants, announced October 15, 2026.", "scope": "NVIDIA's releases provide varied openness and efficiency claims; these are model family claims and do not establish use by a particular customer or military/Israel-related deployment.", "releaseNote": "NVIDIA announced Nemotron 3.5 Super VL and Bolt variants on October 15, 2026. The E4B model was scheduled for October 23; publication/availability of that exact variant was not independently confirmed in the release source.", "dimensions": { "energy": { "summary": "NVIDIA frames Lightning and Bolt around efficiency and local deployment; it publishes throughput/size claims, not model-specific energy, water, or emissions measurements. Faster throughput is not an energy-per-task result.", "sourceUrls": [ "https://blogs.nvidia.com/blog/nemotron-3-5-super-vl/", "https://blogs.nvidia.com/blog/nemotron-lightning-switchyard-rtx-dgx/" ], "confidence": "medium", "unknowns": "No independently measured power, water, or emissions figures for the family.", "suggestedScore": null, "scoreReason": "No score: the available sources report hardware, throughput, or provider-wide efficiency framing, but no model-specific energy, water, or emissions impact. Efficiency claims alone do not establish a lower footprint.", "assessmentType": "conditional", "argument": "NVIDIA’s announcements give throughput, model size, and local deployment claims but no electricity or emissions measurements. Such figures can help estimate capacity but cannot establish energy per useful task without hardware, utilization, and quality measurements.", "score": null, "href": "/models/nemotron-3-5/#energy" }, "training-rights": { "summary": "NVIDIA says some Nemotron releases include open datasets and training recipes; the October Super VL/Bolt announcement emphasizes open models but this page does not enumerate all data rights. Check the exact checkpoint card and license.", "sourceUrls": [ "https://blogs.nvidia.com/blog/nemotron-3-5-super-vl/", "https://blogs.nvidia.com/blog/nemotron-3-super-agentic-ai/" ], "confidence": "medium", "unknowns": "Open data/recipe completeness differs by release; exact Super VL corpus rights need verification.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "NVIDIA says some Nemotron releases include datasets and recipes, while the cited 3.5 announcements do not enumerate the exact family’s corpus rights. Openness varies by release artifact; inspect each checkpoint’s dataset and license rather than generalizing from the family label.", "score": null, "href": "/models/nemotron-3-5/#training-rights" }, "privacy": { "summary": "Open weights permit local deployment, where prompt handling is controlled by the operator; managed NIM/cloud use has separate provider terms. The model release does not define all downstream privacy practices.", "sourceUrls": [ "https://blogs.nvidia.com/blog/nemotron-3-5-super-vl/" ], "confidence": "medium", "unknowns": "Need deployment-specific logging, retention, and data-training terms.", "suggestedScore": 7, "scoreReason": "Open weights support local deployment and operator control, though hosted providers have separate terms.", "assessmentType": "conditional", "argument": "Open weights permit local hosting where the operator controls prompts, while NIM and cloud deployments introduce provider-specific handling. This privacy benefit applies only to deployments where the operator manages storage, telemetry, and access.", "score": 7, "href": "/models/nemotron-3-5/#privacy" }, "openness": { "summary": "NVIDIA describes Super VL and Bolt as open models and says users can post-train and deploy them; licenses, datasets, and degree of full-stack reproducibility should be checked per artifact.", "sourceUrls": [ "https://blogs.nvidia.com/blog/nemotron-3-5-super-vl/", "https://blogs.nvidia.com/blog/nemotron-3-super-agentic-ai/" ], "confidence": "medium", "unknowns": "Not all family releases necessarily share the same license or data openness.", "suggestedScore": 7, "scoreReason": "NVIDIA describes open models and customization; exact artifact license/data completeness varies by release.", "assessmentType": "conditional", "argument": "NVIDIA describes the relevant models as open and supports post-training and deployment, but openness depends on the individual checkpoint’s license and accompanying datasets. The available weights and customization support reuse, while incomplete training data and recipes limit reproduction.", "score": 7, "href": "/models/nemotron-3-5/#openness" }, "labour": { "summary": "NVIDIA’s Nemotron 3.5 Content Safety card says its prompt, image, and response data were human-annotated through SuperAnnotate, with synthetic data and LLM judges also used. It does not publish annotator pay or working conditions.", "sourceUrls": [ "https://huggingface.co/nvidia/Nemotron-3.5-Content-Safety" ], "confidence": "low", "unknowns": "The Content Safety card does not establish pay, working conditions or protections across every Nemotron 3.5 variant. Current vendor-level outcomes remain unverified.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The model card identifies a named human-annotation vendor and says annotators were instructed to skip very graphic images. This gives concrete evidence of human work and one exposure control, while leaving compensation and broader protections unanswered.", "score": null, "href": "/models/nemotron-3-5/#labour" }, "military-contracts": { "summary": "NVIDIA says Palantir uses Nemotron open models for U.S. government agencies and says Nemotron models are available through AWS GovCloud at DoD Impact Levels 2, 4, and 5. These company materials do not name Nemotron 3.5 as a military deployment.", "sourceUrls": [ "https://www.nvidia.com/en-us/industries/government/federal/", "https://blogs.nvidia.com/blog/palantir-secure-ai-us-agencies-nemotron-open-models/" ], "confidence": "medium", "unknowns": "Which Nemotron releases are deployed, whether a specific use is military rather than civilian government work, and the actual mission and safeguards remain unestablished.", "suggestedScore": null, "scoreReason": "The sources establish family-level government use and availability in defense-cleared cloud environments, but do not identify the mission or safeguards of a Nemotron 3.5 military deployment. Hosting eligibility alone is not enough for a numerical judgment.", "assessmentType": "conditional", "argument": "NVIDIA’s federal page documents Nemotron in government deployments and DoD-cleared hosting, and its Palantir announcement names Nemotron open models serving U.S. agencies. The pages do not identify the 3.5 checkpoint or a particular military mission, so they establish family-level government use rather than a 3.5-specific contract.", "assessmentScope": "Company / model family record", "score": null, "href": "/models/nemotron-3-5/#military-contracts" }, "israel-palestine": { "summary": "We did not find NVIDIA among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "NVIDIA is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/nemotron-3-5/#israel-palestine" } }, "sources": [ { "url": "https://blogs.nvidia.com/blog/nemotron-3-5-super-vl/", "title": "Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models | NVIDIA Blog", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://blogs.nvidia.com/blog/nemotron-lightning-switchyard-rtx-dgx/", "title": "Nemotron 3.5 Lightning and NeMo Switchyard", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://blogs.nvidia.com/blog/nemotron-3-super-agentic-ai/", "title": "New NVIDIA Nemotron 3 Super", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://blogs.nvidia.com/blog/open-models-europe-industry/", "title": "NVIDIA open models advance Europe's sovereign intelligence", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/nvidia/Nemotron-3.5-Content-Safety", "title": "NVIDIA Nemotron 3.5 Content Safety model card and dataset description", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.nvidia.com/en-us/industries/government/federal/", "title": "NVIDIA U.S. federal government offerings", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://blogs.nvidia.com/blog/palantir-secure-ai-us-agencies-nemotron-open-models/", "title": "NVIDIA announcement on Palantir deployment of Nemotron for U.S. agencies", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/nemotron-3-5/", "overall": { "score": 7, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 7 } }, { "id": "olmo-3-7b-instruct", "name": "Olmo 3 7B Instruct", "category": "models", "organization": "AI2 (Allen Institute for AI)", "website": "https://allenai.org/olmo", "summary": "Fully open 7B instruction model with published checkpoints, training code, data, and evaluation artifacts. Its open release supports independent inspection and self-hosted privacy controls.", "scope": "Olmo 3 source/training statements refer to the family and model flow. Energy, labour and use relationships are not implied by openness; remaining claims are limited to published project materials.", "releaseNote": "AI2's release documentation identifies Olmo 3 7B Instruct as part of its November 2025 release and publishes model-flow artifacts. Checked 2026-10-23; no newer Olmo language-model release was verified in the official docs used.", "dimensions": { "energy": { "summary": "AI2's official OLMo 3 training README reports 5.93 trillion pretraining tokens for the 7B base model and identifies the 7B Instruct post-training stages/checkpoints. Training scripts disclose accelerator counts for portions of the 7B pipeline, but the checked README/card do not give end-to-end GPU-hours, device energy, facility electricity, emissions, or water. These base-family details are not a complete measured footprint for the final Instruct checkpoint; no standardized inference-energy figure is reported.", "sourceUrls": [ "https://huggingface.co/allenai/Olmo-3-7B-Instruct", "https://github.com/allenai/OLMo-core/blob/main/src/scripts/official/OLMo3/README.md" ], "confidence": "medium", "unknowns": "No complete 7B Instruct pipeline runtime/energy/emissions or standardized inference measurement; accelerator counts alone do not establish electricity use.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "AI2 discloses token volume and parts of the training pipeline, but not end-to-end energy, emissions, or inference use for the final Instruct checkpoint. Those details make the run more reproducible without quantifying its footprint; compare only after a bounded energy report exists.", "score": null, "href": "/models/olmo-3-7b-instruct/#energy" }, "training-rights": { "summary": "AI2 publishes the Dolma 3/Dolci data curriculum and says it follows web standards, excluding sites that explicitly disallow crawling and paywalled content. This is meaningful provenance disclosure but not a complete legal clearance of every source.", "sourceUrls": [ "https://allenai.org/blog/olmo3", "https://allenai.org/olmo" ], "confidence": "high", "unknowns": "No independent rights audit; dataset curation and rights status remain developer-reported.", "suggestedScore": 7, "scoreReason": "Published corpus and rights-handling disclosures are unusually detailed, but remain developer-reported and lack independent rights verification; score reflects transparency with that material limitation.", "assessmentType": "conditional", "argument": "AI2 publishes Dolma 3/Dolci sources and describes excluding sites that disallow crawling or paywalled content. That is a meaningful rights-respecting process, though web crawling rules do not independently settle every copyright claim or source permission.", "score": 7, "href": "/models/olmo-3-7b-instruct/#training-rights" }, "privacy": { "summary": "Weights can be run locally, letting an operator keep inference inputs on their own infrastructure; hosted deployment privacy depends on the operator. This does not establish privacy properties of AI2-hosted demos.", "sourceUrls": [ "https://docs.allenai.org/models/olmo/index.html", "https://allenai.org/olmo" ], "confidence": "medium", "unknowns": "No complete hosted-demo retention policy assessed.", "suggestedScore": 7, "scoreReason": "Public weights enable local deployment and operator control over prompts, but that is a deployment option rather than a guaranteed privacy safeguard; hosted services and downstream configurations require separate review.", "assessmentType": "conditional", "argument": "AI2 offers local weights and deployment guidance, allowing an operator to keep prompts on its own infrastructure. The materials do not address local telemetry or hosted vendors, so privacy depends on the chosen deployment.", "score": 7, "href": "/models/olmo-3-7b-instruct/#privacy" }, "openness": { "summary": "AI2 publishes weights, code, pretraining and post-training datasets, and evaluation methods, and describes the family as fully open. This supports strong reproducibility compared with model-only weight releases.", "sourceUrls": [ "https://allenai.org/olmo", "https://docs.allenai.org/models/olmo/index.html", "https://docs.allenai.org/latest-releases/index.html" ], "confidence": "high", "unknowns": "Check exact checkpoint license and dependencies for a specific deployment.", "suggestedScore": 9, "scoreReason": "Weights, training data, code, evaluation methods, and training stages are published.", "assessmentType": "evidence", "argument": "AI2 releases weights, code, training datasets, and evaluation methods, enabling substantial independent inspection and reproduction. Reproducing a frontier-scale run still requires substantial resources, and some operational dependencies may not be captured in the released materials.", "score": 9, "href": "/models/olmo-3-7b-instruct/#openness" }, "labour": { "summary": "The release describes human preference data in post-training but does not disclose worker count, pay, working conditions, or contractor/vendor chain for Olmo 3.", "sourceUrls": [ "https://allenai.org/blog/olmo3" ], "confidence": "medium", "unknowns": "Human data contribution is described; labour conditions remain undisclosed in the source checked.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The Olmo 3 release identifies human preference data but does not disclose annotator pay, working conditions, or vendor arrangements. That confirms human input while leaving the fairness of its production unknown; request worker sourcing details.", "score": null, "href": "/models/olmo-3-7b-instruct/#labour" }, "military-contracts": { "summary": "Ai2 publishes Olmo-3-7B-Instruct under Apache 2.0 and provides the checkpoint for independent use. That enables downstream deployment without an Ai2 service contract; no cited source establishes military use.", "sourceUrls": [ "https://huggingface.co/allenai/Olmo-3-7B-Instruct" ], "confidence": "low", "unknowns": "No complete downstream-use register for open weights.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "The exact model card gives Apache 2.0 terms, so a third party can run the checkpoint independently of Ai2. This limits what can be inferred from company contracts or hosted-customer disclosures; it does not show that any military organization has adopted Olmo.", "score": null, "href": "/models/olmo-3-7b-instruct/#military-contracts" }, "israel-palestine": { "summary": "We did not find AI2 (Allen Institute for AI) among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "AI2 (Allen Institute for AI) is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/olmo-3-7b-instruct/#israel-palestine" } }, "sources": [ { "url": "https://allenai.org/blog/olmo3", "title": "Olmo 3: Charting a path through the model flow", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://allenai.org/olmo", "title": "Olmo from Ai2", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.allenai.org/models/olmo/index.html", "title": "Olmo model documentation", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.allenai.org/latest-releases/index.html", "title": "Ai2 latest releases", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/allenai/OLMo-core/blob/main/src/scripts/official/OLMo3/README.md", "title": "OLMo 3 official training README", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/allenai/Olmo-3-7B-Instruct", "title": "Olmo 3 7B Instruct model card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/olmo-3-7b-instruct/", "overall": { "score": 7.7, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 3 } }, { "id": "apertus-1-5-8b", "name": "Apertus 1.5 8B", "category": "models", "organization": "Swiss AI Initiative (EPFL, ETH Zurich, CSCS)", "website": "https://www.apertus-ai.org/", "summary": "Apache 2.0 open model with public weights, training data, and methods. This is the current verified 8B family member; Apertus-8B-2509 is an older version retained separately below.", "scope": "Apertus model-card claims are developer statements. Its charter/initiative-level mission does not establish downstream use, energy per query, or worker conditions.", "releaseNote": "Official Apertus documentation lists Apertus 1.5 8B and 70B as current releases, superseding the initial 2025 Apertus 1.0/8B-2509 line. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "The Apertus 1.5 technical report says the 8B model received 4T additional pretraining tokens beyond Apertus 1.0 and added multimodal support. The card and report provide no 1.5-8B GPU-hours, device power, kWh, CO2e, water, or standardized inference-energy figure. Earlier Apertus 1.0 70B compute estimates must not be attributed to this 8B checkpoint. Local serving instructions show deployability, not low energy use.", "sourceUrls": [ "https://huggingface.co/swiss-ai/Apertus-v1.5-8B", "https://arxiv.org/abs/2610.15542" ], "confidence": "high", "unknowns": "No 1.5-8B training energy/emissions or inference energy measurement; the additional 4T tokens describe training scale, not electricity.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The 1.5 report documents additional training tokens but publishes no 8B energy or emissions total. Energy for an earlier 70B run cannot be transferred to this checkpoint; checkpoint-specific compute and carbon data would be needed for a comparison.", "score": null, "href": "/models/apertus-1-5-8b/#energy" }, "training-rights": { "summary": "The Apertus project describes open training data and methods and says it respects data-owner opt-outs, including retrospective opt-outs. These are developer claims; complete documentation and technical report are available, but no independent rights audit was found.", "sourceUrls": [ "https://www.apertus-ai.org/pages/about/", "https://huggingface.co/swiss-ai/Apertus-8B-2509" ], "confidence": "high", "unknowns": "Verify whether the 1.5 release preserves the same opt-out mechanisms and dataset coverage as 1.0.", "suggestedScore": 7, "scoreReason": "Published corpus and rights-handling disclosures are unusually detailed, but remain developer-reported and lack independent rights verification; score reflects transparency with that material limitation.", "assessmentType": "conditional", "argument": "The project says its dataset is open and supports retrospective data-owner opt-outs, and links to training materials. Those are concrete governance commitments, though they remain developer statements and do not constitute an independent rights audit.", "score": 7, "href": "/models/apertus-1-5-8b/#training-rights" }, "privacy": { "summary": "Self-hosted open weights allow prompts to remain with the operator. The earlier 8B-2509 card says no PII output filter was provided and recommends applying a periodically updated filter; that caveat is version-specific and cannot be assumed fixed in 1.5.", "sourceUrls": [ "https://huggingface.co/swiss-ai/Apertus-8B-2509", "https://www.apertus-ai.org/pages/documentation/" ], "confidence": "medium", "unknowns": "Need current 1.5 output-filter status and hosting privacy details.", "suggestedScore": 7, "scoreReason": "Public weights enable local deployment and operator control over prompts, but that is a deployment option rather than a guaranteed privacy safeguard; hosted services and downstream configurations require separate review.", "assessmentType": "conditional", "argument": "Apertus 1.5 can be self-hosted, but the documented absence of an output PII filter in the earlier 8B card shows local control does not automatically prevent personal-data disclosure. Operators should add appropriate filtering and evaluate their own logs and access controls.", "score": 7, "href": "/models/apertus-1-5-8b/#privacy" }, "openness": { "summary": "Official Apertus docs list 1.5 8B and 70B under Apache 2.0; the project describes architecture, weights, training data, and methods as accessible. Check the specific release's full artifacts before claiming full reproducibility.", "sourceUrls": [ "https://www.apertus-ai.org/pages/documentation/", "https://www.apertus-ai.org/pages/about/" ], "confidence": "high", "unknowns": "Confirm complete 1.5 training intermediates and datasets are published.", "suggestedScore": 9, "scoreReason": "Official project describes Apache 2.0 access to weights, data, and methods.", "assessmentType": "evidence", "argument": "The project publishes weights under Apache 2.0 and describes model methods and training data, allowing reuse and inspection. This is strong openness, while an independent audit of data completeness and rights is still absent.", "score": 9, "href": "/models/apertus-1-5-8b/#openness" }, "labour": { "summary": "Disclosure gap: public technical materials reviewed identify research teams and training hardware but do not disclose data annotator pay, contract conditions, or worker protections.", "sourceUrls": [ "https://www.apertus-ai.org/pages/documentation/" ], "confidence": "low", "unknowns": "No workforce or annotation labour disclosure found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Apertus documentation names research teams and infrastructure but does not report data annotator pay or contractor conditions. Research-team attribution is not workforce transparency; ask for details about any human-curated data before drawing a labor conclusion.", "score": null, "href": "/models/apertus-1-5-8b/#labour" }, "military-contracts": { "summary": "Apertus lists its 1.5 8B checkpoint under Apache 2.0 and links downloadable weights. Independent use is possible without a CSCS or EPFL service contract; the project release pages do not document military use.", "sourceUrls": [ "https://www.apertus-ai.org/pages/documentation/", "https://huggingface.co/swiss-ai/Apertus-v1.5-8B" ], "confidence": "low", "unknowns": "Needs specific downstream customer/procurement evidence.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "Apertus’ official documentation labels the 1.5 8B release Apache 2.0 and links directly to its weights. That means downstream users can deploy it without the research partners as a service provider; it is not evidence of any military deployment.", "score": null, "href": "/models/apertus-1-5-8b/#military-contracts" }, "israel-palestine": { "summary": "We did not find Swiss AI Initiative (EPFL, ETH Zurich, CSCS) among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Swiss AI Initiative (EPFL, ETH Zurich, CSCS) is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/apertus-1-5-8b/#israel-palestine" } }, "sources": [ { "url": "https://www.apertus-ai.org/pages/documentation/", "title": "Apertus documentation and current releases", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.apertus-ai.org/pages/about/", "title": "About Apertus", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/swiss-ai/Apertus-8B-2509", "title": "Apertus-8B-2509 model card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.epfl.ch/labs/mlo/", "title": "EPFL Machine Learning and Optimization Laboratory", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/swiss-ai/Apertus-v1.5-8B", "title": "Apertus v1.5 8B model card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://arxiv.org/abs/2610.15542", "title": "Apertus 1.5 technical report", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/apertus-1-5-8b/", "overall": { "score": 7.7, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 1 } }, { "id": "apertus-8b-2509", "name": "Apertus-8B-2509", "category": "models", "organization": "Swiss AI Initiative (EPFL, ETH Zurich, CSCS)", "website": "https://huggingface.co/swiss-ai/Apertus-8B-2509", "summary": "Original 8B 2025 open checkpoint, now superseded by Apertus 1.5 8B. It remains downloadable under Apache 2.0 and has a detailed public training card.", "scope": "Findings are specific to the 2509 checkpoint and its model card; current Apertus 1.5 improvements or safeguards are not attributed to this older release.", "releaseNote": "Swiss AI's EPFL lab page records Apertus 8B/70B model weights released September 2, 2025; official documentation now lists 1.5 as the current generation. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "The 8B model card reports 15T pretraining tokens and full training details, but does not provide an 8B-specific GPU-hour, kWh, emissions, or water total. The technical report's approximately 6 million GPU-hours and estimated 5 GWh at 560 W per module apply to the 70B model's 15T-token main pretraining run, not this 8B checkpoint; the report says CSCS relies entirely on hydropower and uses closed-cycle Lake Lugano cooling with no water consumed. The 70B figures are not assigned to the 8B entry. No standardized inference-energy measurement is reported.", "sourceUrls": [ "https://huggingface.co/swiss-ai/Apertus-8B-2509", "https://arxiv.org/pdf/2509.14233" ], "confidence": "high", "unknowns": "No exact 8B training energy/emissions or inference per-task/per-token measurement; the family report’s 70B estimate is not transferable to this release.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "The 8B card gives token scale and reproducible training details, while the report’s energy and hydropower figures apply to the separate 70B run. This shows training-process transparency, not the 8B checkpoint’s footprint; do not transfer the larger model’s numbers.", "score": null, "href": "/models/apertus-8b-2509/#energy" }, "training-rights": { "summary": "The model card says training data is open and compliant, respects data-owner opt-outs retrospectively, and links to data reconstruction scripts. These are developer-reported controls, not independent legal certification.", "sourceUrls": [ "https://huggingface.co/swiss-ai/Apertus-8B-2509", "https://www.epfl.ch/labs/mlo/" ], "confidence": "high", "unknowns": "No independent audit of every source or rights claim.", "suggestedScore": 7, "scoreReason": "Published corpus and rights-handling disclosures are unusually detailed, but remain developer-reported and lack independent rights verification; score reflects transparency with that material limitation.", "assessmentType": "conditional", "argument": "The card says data are open and honors retrospective opt-outs, with reconstruction scripts available. These practices support a positive process judgment, but developer claims and scripts do not independently verify every rightsholder’s permission.", "score": 7, "href": "/models/apertus-8b-2509/#training-rights" }, "privacy": { "summary": "Self-hosting lets the operator control prompts. The card states no PII output filter was provided at publication and advises users to apply the project's periodically updated filter; it also provides privacy/copyright request contacts.", "sourceUrls": [ "https://huggingface.co/swiss-ai/Apertus-8B-2509" ], "confidence": "high", "unknowns": "No centrally hosted service retention policy applies to all self-hosted uses; operator controls matter.", "suggestedScore": 7, "scoreReason": "Local inference is possible, but the card says no PII output filter was provided at release.", "assessmentType": "conditional", "argument": "Local weights let operators control prompt storage, but the card says no PII output filter was supplied and recommends periodic opt-out updates. Deployment can protect input privacy while still requiring output controls and refreshed data handling.", "score": 7, "href": "/models/apertus-8b-2509/#privacy" }, "openness": { "summary": "Weights are published under Apache 2.0, with open data, training code, recipes, intermediate checkpoints, and training details described in the model card and project materials.", "sourceUrls": [ "https://huggingface.co/swiss-ai/Apertus-8B-2509", "https://www.apertus-ai.org/pages/about/" ], "confidence": "high", "unknowns": "Confirm linked artifacts remain available and match the exact checkpoint revision.", "suggestedScore": 9, "scoreReason": "Weights, data, training code, recipes, and intermediate checkpoints are publicly described.", "assessmentType": "evidence", "argument": "Apache 2.0 weights, training code, recipes, data, and intermediate checkpoints provide unusually broad inspection and reuse. Full reproduction still depends on the linked data and environment, and independent verification of dataset rights is not available.", "score": 9, "href": "/models/apertus-8b-2509/#openness" }, "labour": { "summary": "Disclosure gap: the model card identifies training infrastructure and research contributors but does not disclose annotation worker compensation, labor conditions, or vendor chain.", "sourceUrls": [ "https://huggingface.co/swiss-ai/Apertus-8B-2509" ], "confidence": "low", "unknowns": "No worker-condition disclosure found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The model card identifies infrastructure and research contributors but does not describe annotation workers’ pay or conditions. The sources leave worker conditions unknown; ask about any human-curated corpora or safety work.", "score": null, "href": "/models/apertus-8b-2509/#labour" }, "military-contracts": { "summary": "The Apertus 8B 2509 card identifies a downloadable model release; the project’s published documentation describes Apache 2.0 as the family’s release license. Independent downstream use is possible, but no checked source identifies military adoption.", "sourceUrls": [ "https://www.apertus-ai.org/pages/documentation/", "https://huggingface.co/swiss-ai/Apertus-8B-2509" ], "confidence": "low", "unknowns": "Needs specific downstream customer/procurement evidence.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "Apertus distributes its checkpoint files and documents the family under Apache 2.0, which allows downstream deployment outside the original research organizations. That weakens any inference from vendor contract lists; no cited deployment record establishes military use of this checkpoint.", "score": null, "href": "/models/apertus-8b-2509/#military-contracts" }, "israel-palestine": { "summary": "We did not find Swiss AI Initiative (EPFL, ETH Zurich, CSCS) among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Swiss AI Initiative (EPFL, ETH Zurich, CSCS) is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/apertus-8b-2509/#israel-palestine" } }, "sources": [ { "url": "https://huggingface.co/swiss-ai/Apertus-8B-2509", "title": "Apertus-8B-2509 model card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.epfl.ch/labs/mlo/", "title": "EPFL Machine Learning and Optimization Laboratory", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.apertus-ai.org/pages/documentation/", "title": "Apertus documentation and current releases", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.apertus-ai.org/pages/about/", "title": "About Apertus", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://arxiv.org/pdf/2509.14233", "title": "Apertus 1.0 technical report (family training compute and CSCS energy context)", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/apertus-8b-2509/", "overall": { "score": 7.7, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 2 } }, { "id": "pleias-pico", "name": "Pleias-Pico", "category": "models", "organization": "PleIAs", "website": "https://huggingface.co/PleIAs/Pleias-Pico", "summary": "353M-parameter specialist RAG model under Apache 2.0. The developer says it uses permissively licensed Common Corpus data and publishes the training settings.", "scope": "Evidence is from the developer-maintained model card; claims about rights compliance and copyright output limitations are developer assertions, not independent certification.", "releaseNote": "The model card identifies Pleias-Pico as a specialized 353M model trained on Common Corpus. The page's collection shows the model was initially published in February 2025; no newer Pico checkpoint was verified here. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "Pleias's model card reports training at Jean-Zay with 16 H100 GPUs using Nanotron, a 90,000-step schedule and one epoch over its RAG dataset. It does not state elapsed accelerator-hours, device/facility electricity, power draw, emissions, or water; steps and GPU count are training-scale context, not energy measurement. No standardized inference-energy figure is reported.", "sourceUrls": [ "https://huggingface.co/PleIAs/Pleias-Pico" ], "confidence": "high", "unknowns": "No elapsed job duration/GPU-hours, kWh, CO2e, water, or per-task inference energy disclosed.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The card gives GPU count and training steps, which describe scale but not elapsed accelerator-hours, facility power, emissions, or water. A comparison needs measured energy and serving data.", "score": null, "href": "/models/pleias-pico/#energy" }, "training-rights": { "summary": "PleIAs says Pico is trained only on permissively licensed Common Corpus content, with dataset and training configuration linked. This is a developer claim; no external rights audit is cited on the card.", "sourceUrls": [ "https://huggingface.co/PleIAs/Pleias-Pico" ], "confidence": "high", "unknowns": "Independent verification of each underlying license/source is not established.", "suggestedScore": 7, "scoreReason": "Published corpus and rights-handling disclosures are unusually detailed, but remain developer-reported and lack independent rights verification; score reflects transparency with that material limitation.", "assessmentType": "conditional", "argument": "Pleias says Pico uses permissively licensed Common Corpus material and publishes the dataset and configuration. This is a stronger provenance claim than a generic web-data statement, but it is still developer-reported and should not be treated as a completed independent audit.", "score": 7, "href": "/models/pleias-pico/#training-rights" }, "privacy": { "summary": "Downloadable weights support local inference, so prompt privacy can be controlled by the deployment operator. The card does not establish privacy of hosted third-party RAG applications or endpoints.", "sourceUrls": [ "https://huggingface.co/PleIAs/Pleias-Pico" ], "confidence": "medium", "unknowns": "No hosted service retention or telemetry policy assessed.", "suggestedScore": 7, "scoreReason": "Public weights enable local deployment and operator control over prompts, but that is a deployment option rather than a guaranteed privacy safeguard; hosted services and downstream configurations require separate review.", "assessmentType": "conditional", "argument": "Downloadable weights make local inference possible, so a deployment can keep prompts under operator control. The card does not cover hosted providers or local logging; check the actual serving setup before handling sensitive prompts.", "score": 7, "href": "/models/pleias-pico/#privacy" }, "openness": { "summary": "The model card lists Apache 2.0, public weights, a Common Corpus dataset link, and training settings. This gives users substantial inspection/reuse access, subject to checking the exact dataset and dependency terms.", "sourceUrls": [ "https://huggingface.co/PleIAs/Pleias-Pico" ], "confidence": "high", "unknowns": "Verify associated data/license artifacts and dependencies at the chosen revision.", "suggestedScore": 9, "scoreReason": "Apache 2.0 weights, data link, and training settings are published.", "assessmentType": "evidence", "argument": "The Apache 2.0 license, public weights, dataset link, and training settings allow substantial inspection and reuse. The remaining limitation is that source-level rights and actual reproduction still require checking the linked data and environment.", "score": 9, "href": "/models/pleias-pico/#openness" }, "labour": { "summary": "Disclosure gap: the model card credits compute and technical contributors but does not disclose annotation workers, pay, working conditions, or contractor protections.", "sourceUrls": [ "https://huggingface.co/PleIAs/Pleias-Pico" ], "confidence": "low", "unknowns": "No labor-supply-chain disclosure found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The model card credits compute and technical contributors but does not say whether human data annotators were used or how any were compensated. Technical credits do not establish workforce conditions; ask for labor sourcing details.", "score": null, "href": "/models/pleias-pico/#labour" }, "military-contracts": { "summary": "Pleias-Pico is published with downloadable weights under Apache 2.0 and local inference instructions. A downstream user could deploy it without a Pleias service contract; no cited source identifies military adoption.", "sourceUrls": [ "https://huggingface.co/PleIAs/Pleias-Pico" ], "confidence": "low", "unknowns": "No downstream-use register for open weights.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "The model card exposes the weights under Apache 2.0 and gives local inference options. Independent deployment therefore need not appear in Pleias customer disclosures, but the license itself says nothing about actual military use.", "score": null, "href": "/models/pleias-pico/#military-contracts" }, "israel-palestine": { "summary": "We did not find PleIAs among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "PleIAs is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/pleias-pico/#israel-palestine" } }, "sources": [ { "url": "https://huggingface.co/PleIAs/Pleias-Pico", "title": "Pleias-Pico model card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/collections/PleIAs/common-models", "title": "PleIAs Common Models collection", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/pleias-pico/", "overall": { "score": 7.7, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 4 } }, { "id": "kimi-k3", "name": "Kimi K3", "category": "models", "organization": "Moonshot AI", "website": "https://www.moonshot.ai/", "summary": "Hosted frontier multimodal model with 2.8T parameters and 1M-token context; Kimi's official site and API announcement identify K3 as the current flagship.", "scope": "Release facts are model-specific. A family release announcement does not establish training-data rights or privacy conditions across consumer and API surfaces; those are recorded as gaps where not sourced.", "releaseNote": "Moonshot's official announcement dated July 22, 2026 calls Kimi K3 its most capable model; the company homepage still promotes K3 on October 23, 2026. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "Disclosure gap: no Kimi K3 training energy, inference energy, water, or emissions data was found. The 2.8T parameter headline and token pricing are not energy measurements.", "sourceUrls": [ "https://www.moonshot.ai/", "https://forum.moonshot.ai/t/kimi-k3-is-here-our-most-capable-model/480" ], "confidence": "low", "unknowns": "No model-specific compute/impact report found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Moonshot’s product and forum announcement describe the K3 model but do not quantify its training or inference footprint. A large parameter count is not an energy measurement; compare only after model-bounded electricity and emissions are reported.", "score": null, "href": "/models/kimi-k3/#energy" }, "training-rights": { "summary": "Disclosure gap: the K3 announcement describes capability and architecture but does not itemize training data, licenses, consent, or opt-out mechanisms.", "sourceUrls": [ "https://forum.moonshot.ai/t/kimi-k3-is-here-our-most-capable-model/480" ], "confidence": "low", "unknowns": "Need K3 model card and training-data rights disclosure.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The K3 launch page describes architecture and capabilities but no corpus, licenses, or opt-outs. With no source-level provenance, users cannot evaluate rights risk from the announcement; request a dataset and governance statement.", "score": null, "href": "/models/kimi-k3/#training-rights" }, "privacy": { "summary": "Disclosure gap: the announcement links to Kimi API but does not itself establish prompt retention, training use, human review, or regional processing controls. Consumer and API settings must be checked separately.", "sourceUrls": [ "https://forum.moonshot.ai/t/kimi-k3-is-here-our-most-capable-model/480" ], "confidence": "low", "unknowns": "No current privacy/retention policy was verified in this pass.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The cited launch page points to API access but does not describe prompt retention, model training, or human review. Privacy for a real deployment therefore needs the current API terms and data-processing policy before sensitive use.", "score": null, "href": "/models/kimi-k3/#privacy" }, "openness": { "summary": "K3 is presented through Moonshot's product and API platform; its announcement does not state open weights or a reproducible training release. Treat as hosted/proprietary pending exact license evidence.", "sourceUrls": [ "https://forum.moonshot.ai/t/kimi-k3-is-here-our-most-capable-model/480", "https://www.moonshot.ai/" ], "confidence": "medium", "unknowns": "No public weights or full training recipe found.", "suggestedScore": 2, "scoreReason": "Hosted/API announcement does not state public weights or a complete training release.", "assessmentType": "evidence", "argument": "Moonshot presents K3 through its product/API service and the cited materials do not announce downloadable weights or a reproducible training package. Treat it as hosted access unless an exact, licensed checkpoint can be verified.", "score": 2, "href": "/models/kimi-k3/#openness" }, "labour": { "summary": "Moonshot’s Kimi K3 materials describe model architecture, large-scale training efficiency claims, and the planned technical report, but do not give annotation-worker sourcing, pay, or protections.", "sourceUrls": [ "https://www.kimi.ai/blog/kimi-k3", "https://huggingface.co/moonshotai/Kimi-K3" ], "confidence": "low", "unknowns": "No labor disclosure reviewed.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Moonshot’s K3 technical announcement discusses its training recipe and recommends a large accelerator deployment, yet does not identify data annotators or working conditions. This supports treating the workforce behind the release as undocumented, rather than inferring conditions from the company’s location or model openness.", "score": null, "href": "/models/kimi-k3/#labour" }, "military-contracts": { "summary": "Moonshot publishes Kimi K3 weights under a model-specific license with instructions for independent serving. That creates a downstream route without a Moonshot contract; the reviewed company materials do not document a military use.", "sourceUrls": [ "https://huggingface.co/moonshotai/Kimi-K3" ], "confidence": "low", "unknowns": "Needs a separate procurement and company announcement review.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "The Kimi K3 model card includes downloadable weights and local-serving instructions under Moonshot’s Kimi K3 License. A third party could therefore operate it outside Moonshot’s hosted API, so absence from company customer disclosures would not settle the question; no cited source demonstrates military use.", "score": null, "href": "/models/kimi-k3/#military-contracts" }, "israel-palestine": { "summary": "We did not find Moonshot AI among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Moonshot AI is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/kimi-k3/#israel-palestine" } }, "sources": [ { "url": "https://forum.moonshot.ai/t/kimi-k3-is-here-our-most-capable-model/480", "title": "Kimi K3 is here: our most capable model", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.moonshot.ai/", "title": "Moonshot AI homepage and latest research", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.kimi.ai/blog/kimi-k3", "title": "Moonshot Kimi K3 launch and technical overview", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/moonshotai/Kimi-K3", "title": "Moonshot Kimi K3 model card and license", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/kimi-k3/", "overall": { "score": 2, "rated": 1, "total": 6, "coverage": 0.16666666666666666, "partial": true, "color": "red", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 22 } }, { "id": "glm-5-3", "name": "GLM-5.3 family", "category": "models", "organization": "Z.ai", "website": "https://z.ai/", "summary": "Hosted model family for coding and long-horizon agents. GLM-5.3 and GLM-5.3-Flash are the latest verified family releases; the main 5.3 announcement said weights were planned after safety review, so current weight status should be rechecked per checkpoint.", "scope": "The release discloses model-development and human-in-the-loop details but not worker conditions. Performance/cyber capabilities and announced open weights do not establish downstream use or license scope.", "releaseNote": "Z.ai announced GLM-5.3 on August 14 and GLM-5.3-Flash on August 26, 2026. Its product pages list GLM-5.3 as latest flagship; checked 2026-10-23.", "dimensions": { "energy": { "summary": "GLM-5.3 release describes increased post-training compute and token efficiency but gives no training kWh, water, emissions, or measured inference energy. Token counts and speed do not establish environmental impact.", "sourceUrls": [ "https://z.ai/blog/glm-5.3", "https://z.ai/blog/glm-5.3-flash" ], "confidence": "medium", "unknowns": "No model-specific energy or lifecycle figures found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Z.ai describes additional post-training compute and token efficiency but does not publish electricity, emissions, or water measurements. More efficient token use can coexist with larger total compute; comparing footprints requires training and serving measurements.", "score": null, "href": "/models/glm-5-3/#energy" }, "training-rights": { "summary": "The release describes scaling RL on accumulated task environments and says human-in-the-loop work remains meaningful; it does not itemize pretraining/post-training data sources or rights for GLM-5.3.", "sourceUrls": [ "https://z.ai/blog/glm-5.3", "https://z.ai/blog/glm-5.3-flash" ], "confidence": "medium", "unknowns": "Need exact corpus provenance, licenses, and rights/consent policy.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The release explains RL environments but does not enumerate pretraining or post-training data rights. Human-in-the-loop environment work is not evidence of data permission; request separate provenance for both data and interaction traces.", "score": null, "href": "/models/glm-5-3/#training-rights" }, "privacy": { "summary": "The release offers hosted API/Coding Plan access and future local weights but does not define prompt training use, retention, human access, or data residency. Privacy depends on selected product and eventual self-hosting setup.", "sourceUrls": [ "https://z.ai/blog/glm-5.3", "https://z.ai/blog/glm-5.3-flash" ], "confidence": "low", "unknowns": "Need current endpoint-specific privacy and data-use terms.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The release offers hosted API and coding access but does not specify prompt retention, training use, human access, or residency. A privacy judgment would require the current service policy and product-specific terms; do not infer privacy from future local weights.", "score": null, "href": "/models/glm-5-3/#privacy" }, "openness": { "summary": "Z.ai described planned weight release for GLM-5.3 after safety evaluation; the checked release source did not independently establish final availability or exact license. Do not yet score as open-weight without the artifact/license.", "sourceUrls": [ "https://z.ai/blog/glm-5.3", "https://z.ai/blog/glm-5.3-flash" ], "confidence": "medium", "unknowns": "Verify checkpoint weight publication and license at current official repository.", "suggestedScore": 4, "scoreReason": "The company announced planned open weights, but the checked source did not verify final artifact/license availability.", "assessmentType": "conditional", "argument": "Z.ai announced planned weight release after safety evaluation, but the record does not verify completed publication or exact license. Until the checkpoint and its terms are publicly available, describe access as hosted rather than open-weight.", "score": 4, "href": "/models/glm-5-3/#openness" }, "labour": { "summary": "Z.ai says GLM-5.3's environment generation and verification still require meaningful human-in-the-loop work. It does not disclose worker count, pay, vendor chain, conditions, or protections.", "sourceUrls": [ "https://z.ai/blog/glm-5.3" ], "confidence": "medium", "unknowns": "Human work is acknowledged; labor conditions remain undisclosed.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Z.ai says human-in-the-loop work remains meaningful in environment creation and verification, but provides no worker counts, pay, or vendor chain. Human contribution is documented, but worker treatment is unknown; request labor disclosures.", "score": null, "href": "/models/glm-5-3/#labour" }, "military-contracts": { "summary": "Z.ai publishes GLM-5.3 weights and local-serving instructions under its GLM-5.3 license. Independent deployment is technically possible subject to those terms; the checked sources do not identify a military customer or deployment.", "sourceUrls": [ "https://huggingface.co/zai-org/GLM-5.3" ], "confidence": "low", "unknowns": "Needs procurement and company-source review; no downstream-use register found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "The official Z.ai model card includes model files and local Transformers/vLLM/SGLang serving instructions, but the license is a model-specific GLM-5.3 license rather than an unrestricted open license. Those facts make third-party deployment possible under its terms; they do not establish a military relationship.", "score": null, "href": "/models/glm-5-3/#military-contracts" }, "israel-palestine": { "summary": "We did not find Z.ai among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Z.ai is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/glm-5-3/#israel-palestine" } }, "sources": [ { "url": "https://z.ai/blog/glm-5.3", "title": "GLM-5.3: Frontier Coding with Emergent Cyber Capabilities", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://z.ai/blog/glm-5.3-flash", "title": "GLM-5.3-Flash: Frontier Intelligence, Flash Cost", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://autoclaw.z.ai/models/", "title": "Z.ai GLM model listing", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/zai-org/GLM-5.3", "title": "Z.ai GLM-5.3 model card and local inference instructions", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/glm-5-3/", "overall": { "score": 4, "rated": 1, "total": 6, "coverage": 0.16666666666666666, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 15 } }, { "id": "minimax-m3-1-flash", "name": "MiniMax M3.1 Flash", "category": "models", "organization": "MiniMax", "website": "https://www.minimax.io/", "summary": "Latest verified MiniMax general text model release in the company newsroom, marketed for agentic tasks, coding, computer use, and cybersecurity. M3 remains a separate broader multimodal frontier release.", "scope": "Release status and claims are from MiniMax company/IR pages. Privacy and rights are gaps unless a direct model or service policy is cited; M3 claims do not automatically transfer to M3.1 Flash.", "releaseNote": "MiniMax's investor-relations release list records M3.1 Flash rollout on October 19, 2026 and identifies M3 as a separate June 1 frontier model. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "MiniMax says M3.1 Flash makes inference nearly three times faster and reduces usage costs by 70%, but publishes no energy, water, or emissions per task. Speed/cost claims are not energy measurements.", "sourceUrls": [ "https://ir.minimax.io/news-events/new-releases", "https://www.minimax.io/blog/minimax-m31f" ], "confidence": "medium", "unknowns": "No model-specific training or inference energy footprint found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "MiniMax reports faster inference and lower usage cost, but neither establishes lower electricity per task or total emissions. The efficiency claim is relevant to performance, not a footprint measure; request energy benchmarks on representative workloads.", "score": null, "href": "/models/minimax-m3-1-flash/#energy" }, "training-rights": { "summary": "The public M3 and M2.5 releases describe architectures, RL environments, and some professional contributors, but the M3.1 Flash newsroom listing does not provide an itemized corpus or rights register. These older-family details are not assumed to apply to 3.1 Flash.", "sourceUrls": [ "https://ir.minimax.io/news-events/new-releases", "https://www.minimax.io/blog/minimax-m3", "https://www.minimax.io/blog/minimax-m31f", "https://agent.minimax.io/tools/m3.1-flash" ], "confidence": "low", "unknowns": "Need M3.1 Flash model card and source-level training-rights disclosure.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Earlier MiniMax releases describe contributors and RL environments, but those facts do not establish M3.1 Flash data sources or rights. The current release does not provide a model-specific provenance ledger, so users should request one before relying on training-rights claims.", "score": null, "href": "/models/minimax-m3-1-flash/#training-rights" }, "privacy": { "summary": "The company release list confirms current API availability but does not establish prompt retention, training use, or human review for M3.1 Flash. Check MiniMax API and product privacy terms before use.", "sourceUrls": [ "https://ir.minimax.io/news-events/new-releases" ], "confidence": "low", "unknowns": "No endpoint-specific privacy controls verified in this pass.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The company release list confirms availability but does not set out M3.1 Flash prompt retention or training rules. Before submitting sensitive information, verify current API terms and data handling for the chosen account and region.", "score": null, "href": "/models/minimax-m3-1-flash/#privacy" }, "openness": { "summary": "MiniMax called M3 an open-weight model in June; the October M3.1 Flash announcement does not establish that M3.1 weights are available or share M3's license. Treat exact M3.1 openness as unknown.", "sourceUrls": [ "https://ir.minimax.io/news-events/new-releases", "https://www.minimax.io/blog/minimax-m3", "https://www.minimax.io/blog/minimax-m31f", "https://agent.minimax.io/tools/m3.1-flash" ], "confidence": "medium", "unknowns": "Need exact M3.1 Flash weights and license source.", "suggestedScore": 3, "scoreReason": "Latest Flash release is verified, but its weights and license were not; earlier M3 openness is not transferred to it.", "assessmentType": "conditional", "argument": "MiniMax called the earlier M3 release open-weight, while the M3.1 Flash announcement does not establish that the Flash checkpoint shares those weights or license. Verify the M3.1 Flash checkpoint and license before describing it as open-weight.", "score": 3, "href": "/models/minimax-m3-1-flash/#openness" }, "labour": { "summary": "MiniMax's M2.5 release says senior professionals contributed requirements, feedback, standards, and training data; no M3.1 Flash labor disclosure or worker-condition details were found. M2.5 specifics are not attributed to M3.1 Flash.", "sourceUrls": [ "https://www.minimax.io/news/minimax-m25", "https://ir.minimax.io/news-events/new-releases" ], "confidence": "low", "unknowns": "No current annotator/contractor count, pay, or worker-protection disclosure.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "The M2.5 announcement credits professional contributors, but it does not establish conditions for M3.1 Flash annotation or evaluation workers. Credit for expertise is not evidence of fair employment; seek current vendor and compensation details.", "score": null, "href": "/models/minimax-m3-1-flash/#labour" }, "military-contracts": { "summary": "MiniMax’s October 19 release says M3.1 Flash is available through its API, subscription plan, and MiniMax Code, with weights to be released later. The checked release does not identify a military customer; a hosted-service contract could exist without public model-specific disclosure.", "sourceUrls": [ "https://www.minimax.io/blog/minimax-m31f" ], "confidence": "low", "unknowns": "Needs procurement and company-source review.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "At the October 19 release checkpoint, MiniMax described M3.1 Flash as an API and product offering and said weights would come later. This means deployment currently depends on MiniMax’s hosted or product channels, but the release is not a customer register and does not show whether defense customers use it.", "score": null, "href": "/models/minimax-m3-1-flash/#military-contracts" }, "israel-palestine": { "summary": "We did not find MiniMax among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "MiniMax is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/minimax-m3-1-flash/#israel-palestine" } }, "sources": [ { "url": "https://ir.minimax.io/news-events/new-releases", "title": "MiniMax new releases", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.minimax.io/blog/minimax-m3", "title": "MiniMax M3: Frontier Coding, 1M Context, Native Multimodality", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.minimax.io/news/minimax-m25", "title": "MiniMax M2.5: Built for Real-World Productivity", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.minimax.io/blog/minimax-m31f", "title": "MiniMax M3.1 Flash: Redefining Agent Performance and Cost", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://agent.minimax.io/tools/m3.1-flash", "title": "MiniMax M3.1 Flash is Now Generally Available", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/minimax-m3-1-flash/", "overall": { "score": 3, "rated": 1, "total": 6, "coverage": 0.16666666666666666, "partial": true, "color": "red", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 21 } }, { "id": "reflection-beam", "name": "Reflection Beam", "category": "models", "organization": "Reflection AI", "website": "https://reflection.ai/", "summary": "501B-parameter, 23B-active open-weight reasoning and coding model; weights and technical report released October 16, 2026 under Apache 2.0.", "scope": "Training and release facts are Beam-specific and unusually detailed. Privacy after self-hosting is controlled by deployment operator; Reflection API terms remain separate. Open weights increase deployment control but also leave downstream uses less observable.", "releaseNote": "Reflection announced Beam October 5 and released weights, model card, and safety report on October 16, 2026. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "Reflection reports 6,144 GB300 GPUs for under four weeks of pretraining and 10.5K GB300 GPUs for a four-week RL run, but no kWh, water, or emissions totals. Compute disclosures are unusually detailed but not environmental accounting.", "sourceUrls": [ "https://reflection.ai/blog/introducing-beam", "https://reflection.ai/blog/beam-is-available" ], "confidence": "high", "unknowns": "No measured training energy, water, or carbon total; no inference energy per task.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "Reflection reports accelerator counts and durations for pretraining and RL, which make compute scale more visible than most releases but still omit facility energy, water, and emissions. Treat these as useful inputs for a future estimate, not as a verified low- or high-carbon footprint.", "score": null, "href": "/models/reflection-beam/#energy" }, "training-rights": { "summary": "Reflection says Beam used 23.8T tokens from web, public sources, and proprietary licensed datasets, and describes filtering/curation. It says almost all publicly accessible and unrestrictively licensed code/docs were included; the page does not list all sources or provide an external rights audit.", "sourceUrls": [ "https://reflection.ai/blog/introducing-beam", "https://reflection.ai/blog/beam-is-available", "https://reflection.ai/beam-model-card", "https://reflection.ai/beam-safety-report" ], "confidence": "medium", "unknowns": "No itemized training corpus, full license manifest, or independent rights audit.", "suggestedScore": 6, "scoreReason": "Reflection discloses public/web and licensed data categories, but no itemized corpus or independent rights audit.", "assessmentType": "conditional", "argument": "Reflection describes a 23.8-trillion-token mixture, public and proprietary licensed sources, filtering, and curation, with further detail in the model card and safety report. This supports a comparatively specific provenance account, but independent confirmation of licenses and coverage remains unavailable.", "score": 6, "href": "/models/reflection-beam/#training-rights" }, "privacy": { "summary": "Beam can be self-hosted under open weights, allowing operators to keep prompts locally. Reflection also offers a hosted API; API privacy and retention need to be assessed separately. No single policy covers all partner endpoints.", "sourceUrls": [ "https://reflection.ai/blog/beam-is-available" ], "confidence": "medium", "unknowns": "Need current first-party API terms and partner-specific logging/retention controls.", "suggestedScore": 7, "scoreReason": "Public weights enable local deployment and operator control over prompts, but that is a deployment option rather than a guaranteed privacy safeguard; hosted services and downstream configurations require separate review.", "assessmentType": "conditional", "argument": "Beam’s weights can be self-hosted, while Reflection also offers an API; those routes have different data controllers and retention risks. Use the local option only with operator controls, and review the API terms separately before sending sensitive prompts.", "score": 7, "href": "/models/reflection-beam/#privacy" }, "openness": { "summary": "Reflection released model weights under Apache 2.0 along with model card, safety technical report, serving formats, and deployment/customization artifacts. This is strong weight and technical openness; source-level training data remains only partially disclosed.", "sourceUrls": [ "https://reflection.ai/blog/beam-is-available", "https://reflection.ai/blog/introducing-beam", "https://reflection.ai/beam-model-card", "https://reflection.ai/beam-safety-report" ], "confidence": "high", "unknowns": "Full training data is not published, so the full training pipeline is not reproducible from public artifacts alone.", "suggestedScore": 9, "scoreReason": "Apache 2.0 weights, model card, safety report, and multiple deployment artifacts are released.", "assessmentType": "evidence", "argument": "Reflection released Apache 2.0 weights plus a model card, safety report, and serving artifacts, supporting substantial reuse and inspection. The training data are described rather than fully reproducible from a public source-by-source corpus, so openness is strong but not complete.", "score": 9, "href": "/models/reflection-beam/#openness" }, "labour": { "summary": "Reflection describes human-in-the-loop curation, environments, grading, and expert review but does not disclose data worker count, compensation, vendor chain, or protections.", "sourceUrls": [ "https://reflection.ai/blog/introducing-beam" ], "confidence": "medium", "unknowns": "Human involvement is described; labor conditions remain undisclosed.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "argument", "argument": "Reflection describes human curation, grading, and expert review but reports no worker counts, compensation, vendors, or protections. The evidence confirms human work but does not describe worker conditions; request details about the labor chain.", "score": null, "href": "/models/reflection-beam/#labour" }, "military-contracts": { "summary": "Reflection says Beam is available to governments and supports sovereign deployments, but the cited launch page does not establish a military contract or military deployment. No model-specific contract was verified in this pass.", "sourceUrls": [ "https://reflection.ai/blog/beam-is-available" ], "confidence": "medium", "unknowns": "No exact defense procurement or end-use evidence found.", "suggestedScore": null, "scoreReason": "No score: the checked primary sources do not provide enough evidence for a defensible editorial rating; this is a disclosure gap, not a positive or negative finding.", "assessmentType": "conditional", "argument": "Reflection says Beam can serve government and sovereign customers, but the cited announcement does not establish military procurement or deployment. Treat public-sector availability as a possible use context, not a defense contract; no military deployment is established.", "score": null, "href": "/models/reflection-beam/#military-contracts" }, "israel-palestine": { "summary": "We did not find Reflection AI among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Reflection AI is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/models/reflection-beam/#israel-palestine" } }, "sources": [ { "url": "https://reflection.ai/blog/introducing-beam", "title": "Introducing Beam: Reflection's 501B open-weight model", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://reflection.ai/blog/beam-is-available", "title": "Beam is now available", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://reflection.ai/beam-model-card", "title": "Beam Model Card | Reflection", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://reflection.ai/beam-safety-report", "title": "Beam Safety Technical Report | Reflection", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/reflection-beam/", "overall": { "score": 7.3, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 5 } }, { "id": "gemma", "name": "Gemma family", "category": "models", "organization": "Google DeepMind / Google", "website": "https://ai.google.dev/gemma/", "summary": "Open-weight family of general-purpose Google models, including Gemma 4 multimodal and text variants.", "scope": "This entry covers the general-purpose Gemma family, especially the current Gemma 4 core releases. Organization-wide privacy, labour, energy, military, and Israel-Palestine relationships are not inferred from this model entry. Specialized Gemma releases are noted separately from the core general-purpose family.", "releaseNote": "The official releases catalog lists Gemma 4 (E2B, E4B, 12B, 26B A4B, and 31B) and a June 3, 2026 Gemma 4 12B Unified release. The catalog also lists EmbeddingGemma 2 on October 6, 2026; that is a specialized embedding release, not a newer general-purpose core checkpoint. Checked 2026-10-23.", "dimensions": { "energy": { "summary": "Disclosure gap: the Gemma 4 card describes small-model and on-device efficiency, but gives no training electricity, GPU-hours, emissions, water, or standardized per-inference energy figures. Hardware footprint claims are not measured energy.", "sourceUrls": [ "https://ai.google.dev/gemma/docs/core/model_card_4", "https://arxiv.org/abs/2607.02770" ], "confidence": "low", "unknowns": "No measured training or inference energy, emissions, or water disclosure was found in the checked Gemma 4 card or technical report.", "suggestedScore": null, "scoreReason": "No defensible score without comparable training or inference impact data.", "assessmentType": "argument", "argument": "The Gemma 4 card’s on-device efficiency claims describe deployment capability rather than measured training or request energy. Without energy, emissions, and water boundaries, the impact cannot be compared; measure the specific local or hosted workload.", "score": null, "href": "/models/gemma/#energy" }, "training-rights": { "summary": "The model card describes a broad mixture of web documents, code, mathematics, images, and audio, with a January 2025 data cutoff and filtering steps. It does not provide a source-level training-data manifest or a legal rights audit.", "sourceUrls": [ "https://ai.google.dev/gemma/docs/core/model_card_4", "https://arxiv.org/abs/2607.02770" ], "confidence": "medium", "unknowns": "Corpus composition by source, consent and licensing status, creator opt-outs, and independent rights review remain undisclosed.", "suggestedScore": 4, "scoreReason": "Some dataset and filtering description is published, but source attribution and rights clearance are not documented.", "assessmentType": "conditional", "argument": "The Gemma 4 card identifies source categories, modalities, filtering, and a January 2025 cutoff, but no source-level manifest or rights audit. Users can understand the broad dataset scope, but cannot verify individual permissions or creator opt-out coverage.", "score": 4, "href": "/models/gemma/#training-rights" }, "privacy": { "summary": "Gemma 4 weights can be downloaded and run locally or on-device, allowing an operator to keep prompts on infrastructure they control. This is a deployment option, not a model-level privacy guarantee; hosted Google services have their own data handling terms.", "sourceUrls": [ "https://ai.google.dev/gemma/docs/core/model_card_4", "https://ai.google.dev/gemma/apache_2" ], "confidence": "medium", "unknowns": "Privacy depends on the chosen runtime, host, telemetry, and application; the model card does not audit deployments.", "suggestedScore": 7, "scoreReason": "Local deployment can reduce data disclosure to a service provider, with safeguards dependent on operator choices.", "assessmentType": "conditional", "argument": "Downloadable Gemma weights can run on-device or on infrastructure controlled by the operator, reducing exposure to a hosted service when configured locally. This does not remove local logging or application risks; hosted Google use requires separate service-policy review.", "score": 7, "href": "/models/gemma/#privacy" }, "openness": { "summary": "Gemma 4 provides downloadable pre-trained and instruction-tuned weights under Apache 2.0, with model cards and a technical report. The training corpus itself is not released as a source-level dataset.", "sourceUrls": [ "https://ai.google.dev/gemma/docs/core/model_card_4", "https://ai.google.dev/gemma/apache_2", "https://arxiv.org/abs/2607.02770" ], "confidence": "high", "unknowns": "Training data, full data processing pipeline, and a fully reproducible training run are not publicly supplied.", "suggestedScore": 8, "scoreReason": "Permissive license, weights, and technical documentation are available; source training data remain closed.", "assessmentType": "evidence", "argument": "Apache 2.0 weights, model cards, and a technical report give broad access to deployment and technical inspection. Because the training corpus is not released, the family is not fully reproducible despite its permissive model license.", "score": 8, "href": "/models/gemma/#openness" }, "labour": { "summary": "Company context: Google DeepMind / Google. Google’s supplier code applies labor and human-rights requirements across direct, temporary, migrant, contract, and subcontracted workers; it allows audits, requires remediation, and sets a 60-hour weekly limit with stated exceptions. This is a supplier policy, not an audit of AI data workers. This is a record of the company behind Gemma family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://about.google/company-info/supplier-code-of-conduct/", "https://ai.google.dev/gemma/docs/core/model_card_4" ], "confidence": "high", "unknowns": "Working conditions for the people building or supporting Gemma family are not established by this company-wide record. No current AI-annotation-specific outcomes, public audit findings, or remediation data for Vertex-related data work were established.", "suggestedScore": 4, "scoreReason": "Company-level assessment: Google publishes broad supplier protections and audit rights, but this review found no worker-level results for AI data work. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "Google publishes broad supplier protections and audit rights, but this review found no worker-level results for AI data work. This is relevant when deciding whose business to support, while pay and working conditions for the team behind Gemma family remain unestablished. Ask for current worker outcomes for that team and its suppliers.", "assessmentScope": "Company record", "productArgument": "The Gemma card reports human safety evaluations but not who performed them, how many workers were involved, or their terms. Safety evaluation evidence does not establish fair labor conditions; request contractor and evaluator disclosures.", "score": 4, "href": "/models/gemma/#labour" }, "military-contracts": { "summary": "Company context: Google DeepMind / Google. Google documents access for Department of War personnel and classified defense workflows. Separately, Israel’s procurement administration identifies Google as a Nimbus public-cloud provider; AFSC and Who Profits describe military/security use, while the Guardian reports contract terms limiting service withdrawal. These sources do not establish Google’s role in a particular attack. This is a record of the company behind Gemma family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://cloud.google.com/blog/topics/public-sector/announcements-and-highlights-from-google-public-sector-summit-2026", "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/alphabet", "https://ai.google.dev/gemma/docs/core/model_card_4", "https://ai.google.dev/gemma/prohibited_use_policy" ], "confidence": "high", "unknowns": "Deployment of Gemma family under these contracts is not established. The source does not disclose contract terms or evaluate downstream deployment outcomes.", "suggestedScore": 3, "scoreReason": "Company-level assessment: Google documents U.S. Department of War and classified defense workflows; Israel’s procurement record identifies Google as a Nimbus cloud provider, and independent research/reporting describes military or security use and limits on withdrawal. The reviewed evidence does not attribute a specific harmful operation to Google. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "conditional", "argument": "Google documents U.S. Department of War and classified defense workflows; Israel’s procurement record identifies Google as a Nimbus cloud provider, and independent research/reporting describes military or security use and limits on withdrawal. The reviewed evidence does not attribute a specific harmful operation to Google. This is relevant when deciding whose business to support. Deployment of Gemma family or this particular model release under those contracts remains unestablished; consider the company relationship separately from the product’s actual use.", "assessmentScope": "Company record", "productArgument": "The Gemma card and prohibited-use policy do not identify a Gemma-specific military contract or deployment, and a use policy cannot prove what customers actually do. No model-specific relationship is established; assess any separate Google relationship only at its documented scope.", "score": 3, "href": "/models/gemma/#military-contracts" }, "israel-palestine": { "summary": "Google / Alphabet is named as a pressure target in the BNC’s official campaign sources. The product is developed or supplied by Google / Alphabet.", "argument": "If following this BNC campaign is a requirement, consider alternatives to paid Google / Alphabet products where practical. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well. This label follows the named company’s campaign status; it does not claim that BDS separately reviewed this exact product.", "assessmentType": "evidence", "confidence": "high", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "pressure", "statusLabel": "Pressure target", "reviewedAt": "2026-10-23", "assessmentScope": "Developer / service company", "bdsMatch": { "target": "Google / Alphabet", "status": "pressure", "scope": "Developer / service company", "relationship": "The product is developed or supplied by Google / Alphabet.", "application": "If following this BNC campaign is a requirement, consider alternatives to paid Google / Alphabet products where practical.", "sourceIds": [] }, "upstreamExposure": null, "businessContext": { "summary": "Company context: Google DeepMind / Google. Israel’s procurement administration says Google was selected with AWS for Nimbus public-cloud services and lists the Ministry of Defense and IDF among project partners. The Guardian’s investigation of leaked agreement documents reports that Nimbus terms limited Google’s ability to suspend service, including after alleged policy violations. Who Profits and AFSC describe cloud and AI use by military/security bodies, including surveillance-related capabilities; these are attributed research claims, not proof of Google’s role in a specific attack. Both companies denied circumventing legal obligations in response to the Guardian investigation. This is a record of the company behind Gemma family; it does not establish use of this particular product or release.", "unknowns": "How Gemma family is used in this relationship is not established. The government record establishes the contract and project participants; the investigative reporting describes contract controls, and research organizations make broader use claims. The reviewed evidence does not establish Google’s contribution to a particular operation, injury, or other specific harm.", "sourceUrls": [ "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/alphabet" ] }, "score": null, "href": "/models/gemma/#israel-palestine" } }, "sources": [ { "url": "https://ai.google.dev/gemma/docs/releases?hl=en", "title": "Gemma releases", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://ai.google.dev/gemma/docs/core/model_card_4", "title": "Gemma 4 model card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://ai.google.dev/gemma/apache_2", "title": "Gemma Terms of Use / Apache 2.0 license", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://arxiv.org/abs/2607.02770", "title": "Gemma 4 technical report", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://ai.google.dev/gemma/prohibited_use_policy", "title": "Gemma prohibited use policy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://about.google/company-info/supplier-code-of-conduct/", "title": "Google Supplier Code of Conduct", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cloud.google.com/blog/topics/public-sector/announcements-and-highlights-from-google-public-sector-summit-2026", "title": "Announcements and highlights from Google Public Sector Summit 2026", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://mr.gov.il/ilgstorefront/en/news/details/111222", "title": "Nimbus Tender — Israel Government Procurement Administration", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "title": "Revealed: Israel demanded Google and Amazon use secret “wink” to sidestep legal orders", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://www.whoprofits.org/publications/report/164", "title": "Digitizing Occupation: The Role of Big Tech in Israeli Military Infrastructure", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://investigate.afsc.org/company/alphabet", "title": "Alphabet / Google — AFSC Investigate", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/models/gemma/", "overall": { "score": 5.2, "rated": 5, "total": 6, "coverage": 0.8333333333333334, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 11 } }, { "id": "gpt-oss", "name": "gpt-oss family", "category": "models", "organization": "OpenAI", "website": "https://openai.com/index/introducing-gpt-oss/", "summary": "Open-weight general-purpose reasoning model family released for self-hosting and third-party deployment.", "scope": "This entry covers gpt-oss-120b and gpt-oss-20b core reasoning models. OpenAI separately describes gpt-oss-safeguard as research-preview safety reasoning models. Privacy depends on the chosen host; OpenAI company policies and contracts are not automatically attributed to these exact weights.", "releaseNote": "OpenAI’s current open-weight model page, updated October 19, 2026, lists gpt-oss-120b and gpt-oss-20b; the release announcement dates the family to August 5, 2025. The checked catalog provides no newer core gpt-oss family release.", "dimensions": { "energy": { "summary": "OpenAI reports training compute of 2.1 million H100-hours for gpt-oss-120b and nearly ten times fewer hours for gpt-oss-20b. GPU-hours are not facility electricity or emissions; the checked sources give no measured kWh, water, or standardized inference-energy figure.", "sourceUrls": [ "https://deploymentsafety.openai.com/gpt-oss" ], "confidence": "high", "unknowns": "No measured electricity, carbon intensity, water, or per-request inference energy is reported.", "suggestedScore": null, "scoreReason": "Training compute is disclosed but cannot be converted to comparable energy or emissions from the provided evidence.", "assessmentType": "conditional", "argument": "OpenAI reports H100 training-hours for both sizes, which gives an auditable compute quantity but not electricity or emissions. Hardware utilization and data-center energy boundaries are still needed; accelerator-hours alone cannot be converted to kWh.", "score": null, "href": "/models/gpt-oss/#energy" }, "training-rights": { "summary": "OpenAI describes training on mostly English text covering STEM, coding, and general knowledge, but does not provide a source-level training-corpus manifest or independent rights audit. The Apache 2.0 license applies to the weights and does not establish training-data rights.", "sourceUrls": [ "https://openai.com/index/introducing-gpt-oss/", "https://openai.com/index/gpt-oss-model-card/" ], "confidence": "medium", "unknowns": "Source-level corpus, consent/licensing evidence, and independent rights review were not found.", "suggestedScore": 4, "scoreReason": "Broad training-data categories are described, while provenance and rights clearance remain opaque.", "assessmentType": "conditional", "argument": "OpenAI describes broad training categories but not a source-level corpus or rights audit; Apache 2.0 applies to the weights, not the training data. These materials do not establish permission for every source, so users should request provenance before making rights claims.", "score": 4, "href": "/models/gpt-oss/#training-rights" }, "privacy": { "summary": "OpenAI says the weights can be run on user-managed or hosting-provider infrastructure and that it does not receive self-hosted inputs unless they are explicitly shared. Privacy therefore depends on the operator and any managed host.", "sourceUrls": [ "https://help.openai.com/en/articles/11870455-openai-open-weight-models-gpt-oss", "https://huggingface.co/openai/gpt-oss-120b" ], "confidence": "high", "unknowns": "The model does not itself enforce retention, access controls, telemetry practices, or host-side data handling.", "suggestedScore": 8, "scoreReason": "Self-hosting can keep prompts outside OpenAI’s service, although the operator’s safeguards determine actual privacy.", "assessmentType": "conditional", "argument": "Self-hosted weights can keep prompt data outside OpenAI’s service, according to its deployment guidance, but the operator then controls logging and access. Self-hosting can protect prompts from OpenAI’s service, but only with sound local controls; managed hosting needs a separate contract review.", "score": 8, "href": "/models/gpt-oss/#privacy" }, "openness": { "summary": "The gpt-oss-120b and 20b weights are downloadable under Apache 2.0, with model cards and safety documentation. OpenAI also describes gpt-oss-safeguard research-preview variants. Training data and the full training pipeline are not released.", "sourceUrls": [ "https://help.openai.com/en/articles/11870455-openai-open-weight-models-gpt-oss", "https://openai.com/index/introducing-gpt-oss/", "https://openai.com/index/gpt-oss-model-card/", "https://huggingface.co/openai/gpt-oss-120b" ], "confidence": "high", "unknowns": "No source-level training corpus or fully reproducible training pipeline is public; open weights do not imply open surrounding services.", "suggestedScore": 8, "scoreReason": "Permissively licensed weights and substantial documentation are available, with training-data and pipeline limits.", "assessmentType": "evidence", "argument": "Downloadable Apache 2.0 weights and public model/safety cards support broad adaptation and inspection. The missing source corpus and full training pipeline prevent full reproducibility, so openness is substantial but bounded.", "score": 8, "href": "/models/gpt-oss/#openness" }, "labour": { "summary": "Company context: OpenAI. OpenAI’s supplier code sets broad labor standards but does not publish worker outcomes. TIME’s investigation reviewed OpenAI–Sama contracts, payslips, and interviews about 2021–22 Kenyan toxic-content labeling; workers described low take-home pay and distress, while Sama disputed workload/counseling claims and OpenAI said Sama managed pay and support. TIME later reported OpenAI said it was unaware of the conditions and no longer worked with Sama; this is a historical contractor case, not evidence of current API annotation conditions. This is a record of the company behind gpt-oss family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://openai.com/policies/supplier-code/", "https://time.com/6247678/openai-chatgpt-kenya-workers/", "https://time.com/7012787/mophat-okinyi/", "https://openai.com/index/introducing-gpt-oss/", "https://openai.com/index/gpt-oss-model-card/" ], "confidence": "medium", "unknowns": "Working conditions for the people building or supporting gpt-oss family are not established by this company-wide record. The sources do not establish whether OpenAI uses other current annotation vendors, present worker outcomes, or the effectiveness of the supplier code across vendors.", "suggestedScore": 4, "scoreReason": "Company-level assessment: A documented past supplier episode includes worker reports of low pay and psychological distress, countered by Sama and OpenAI responses; the current supplier code has no public independent outcome audit. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "A documented past supplier episode includes worker reports of low pay and psychological distress, countered by Sama and OpenAI responses; the current supplier code has no public independent outcome audit. This is relevant when deciding whose business to support, while pay and working conditions for the team behind gpt-oss family remain unestablished. Ask for current worker outcomes for that team and its suppliers.", "assessmentScope": "Company record", "productArgument": "The release and model card do not report annotation workers’ pay, contractors, or working conditions. Safety documentation and model weights do not answer labor questions; seek workforce sourcing details before drawing a labor conclusion.", "score": 4, "href": "/models/gpt-oss/#labour" }, "military-contracts": { "summary": "Company context: OpenAI. OpenAI publicly describes a 2026 classified Department of War agreement with cloud-only deployment, retained safety controls, cleared staff, and contractual limits. It separately says ChatGPT will reach GenAI.mil for unclassified work used by 3 million civilian and military personnel. Both are direct defense deployments; limits are company-reported and not independently audited. This is a record of the company behind gpt-oss family; it does not establish use of this particular product or release.", "sourceUrls": [ "https://openai.com/index/our-agreement-with-the-department-of-war/", "https://openai.com/index/bringing-chatgpt-to-genaimil/", "https://help.openai.com/en/articles/11870455-openai-open-weight-models-gpt-oss" ], "confidence": "high", "unknowns": "Deployment of gpt-oss family under these contracts is not established. Actual use, oversight effectiveness, and compliance with contract limits are not independently established.", "suggestedScore": 5, "scoreReason": "Company-level assessment: OpenAI reports cloud-only deployment, retained safety controls, cleared staff, and contractual limits; these are meaningful safeguards, but no independent check of enforcement is public. This rating describes the company relationship or supplier policy, not a finding about this particular product’s use.", "assessmentType": "evidence", "argument": "OpenAI reports cloud-only deployment, retained safety controls, cleared staff, and contractual limits; these are meaningful safeguards, but no independent check of enforcement is public. This is relevant when deciding whose business to support. Deployment of gpt-oss family or this particular model release under those contracts remains unestablished; consider the company relationship separately from the product’s actual use.", "assessmentScope": "Company record", "productArgument": "OpenAI’s Department of War agreement is a company-level relationship, while the checked public description does not name gpt-oss. The agreement does not establish military use of these downloadable weights; users with defense restrictions should seek an explicit model-specific deployment statement.", "score": 5, "href": "/models/gpt-oss/#military-contracts" }, "israel-palestine": { "summary": "We did not find OpenAI among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "OpenAI is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. OpenAI’s April 2026 agreement names Microsoft as its primary cloud partner and a major shareholder, with revenue-share payments to Microsoft through 2030. That is a documented commercial link to a BDS target, not a separate official listing of OpenAI; local use of downloaded weights does not by itself establish a new payment to either company.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://openai.com/index/next-phase-of-microsoft-partnership/" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": { "label": "Microsoft partnership", "sourceIds": [ "openai-ms" ], "argument": "OpenAI’s April 2026 agreement names Microsoft as its primary cloud partner and a major shareholder, with revenue-share payments to Microsoft through 2030. That is a documented commercial link to a BDS target, not a separate official listing of OpenAI; local use of downloaded weights does not by itself establish a new payment to either company." }, "businessContext": null, "score": null, "href": "/models/gpt-oss/#israel-palestine" } }, "sources": [ { "url": "https://help.openai.com/en/articles/11870455-openai-open-weight-models-gpt-oss", "title": "OpenAI open-weight models: gpt-oss", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/introducing-gpt-oss/", "title": "Introducing gpt-oss", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/gpt-oss-model-card/", "title": "gpt-oss model card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/openai/gpt-oss-120b", "title": "OpenAI gpt-oss-120b model card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://deploymentsafety.openai.com/gpt-oss", "title": "OpenAI deployment safety: gpt-oss", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/our-agreement-with-the-department-of-war/", "title": "Our agreement with the Department of War", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/policies/supplier-code/", "title": "OpenAI Supplier Code of Conduct", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://time.com/6247678/openai-chatgpt-kenya-workers/", "title": "OpenAI used Kenyan workers on less than $2 per hour: TIME investigation", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://time.com/7012787/mophat-okinyi/", "title": "Mophat Okinyi: TIME100 AI 2024", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/bringing-chatgpt-to-genaimil/", "title": "Bringing ChatGPT to GenAI.mil", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/next-phase-of-microsoft-partnership/", "title": "OpenAI: amended Microsoft partnership, April 27 2026", "kind": "primary", "checkedAt": "2026-10-23" } ], "href": "/models/gpt-oss/", "overall": { "score": 5.8, "rated": 5, "total": 6, "coverage": 0.8333333333333334, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 8 } }, { "id": "openai-api", "name": "OpenAI API", "category": "providers", "organization": "OpenAI, L.L.C.", "website": "https://openai.com/api/", "summary": "Direct API access to OpenAI models. Customer API data controls are distinct from consumer ChatGPT terms and from underlying training-corpus rights.", "scope": "OpenAI is both model developer and API operator; this record covers the API service and corporate relationships where expressly documented.", "dimensions": { "energy": { "summary": "OpenAI does not publish a per-request API energy, water, or emissions figure in the reviewed sources. No vendor-specific inference-route footprint established here. OpenAI says it contracted for data-center capacity and describes expected power, cooling, jobs and plans; project-specific statements, not per-request/API footprint.", "sourceUrls": [ "https://openai.com/index/openai-joins-ports-pike-project/" ], "confidence": "medium", "unknowns": "Corporate energy procurement, facility mix, model size, utilization, region, and request-level energy remain unquantified.", "suggestedScore": 4, "scoreReason": "The company discloses large project-level data-center capacity and energy needs, but no region/request footprint; the high-load project context raises concern without quantifying API emissions.", "assessmentType": "conditional", "argument": "The company discloses large project-level data-center capacity and energy needs, but no region/request footprint; the high-load project context raises concern without quantifying API emissions. For a deployment, request or export the selected service and region’s usage-linked footprint when available; do not treat company-wide renewable matching as per-request impact.", "score": 4, "href": "/providers/openai-api/#energy" }, "training-rights": { "summary": "The API data-control documentation says API inputs are not used to train or improve OpenAI models unless the customer opts in. This describes customer-submitted API data; it does not disclose the rights or provenance of the pretraining corpus.", "sourceUrls": [ "https://developers.openai.com/api/docs/guides/your-data" ], "confidence": "high", "unknowns": "Corpus sources, licenses, consent, compensation, and dataset-level exceptions are not established by this API policy.", "suggestedScore": null, "scoreReason": "API opt-out governs customer prompts and outputs, not provenance or licensing of the model pretraining corpus; no corpus-rights due-diligence result was established.", "assessmentType": "conditional", "argument": "The API data-control documentation says API inputs are not used to train or improve OpenAI models unless the customer opts in. This describes customer-submitted API data; it does not disclose the rights or provenance of the pretraining corpus. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/openai-api/#training-rights" }, "privacy": { "summary": "API content is not used for training by default; abuse-monitoring logs may retain content up to 30 days, with feature state and eligible retention controls varying by API feature/account. The cited general privacy policy expressly excludes business API data.", "sourceUrls": [ "https://developers.openai.com/api/docs/guides/your-data" ], "confidence": "high", "unknowns": "Actual storage depends on endpoint, feature, account eligibility/settings, legal retention, and configuration.", "suggestedScore": 7, "scoreReason": "API data is excluded from training by default and retention controls exist, but abuse-monitoring retention can reach 30 days and exceptions vary by feature/account.", "assessmentType": "conditional", "argument": "API data is excluded from training by default and retention controls exist, but abuse-monitoring retention can reach 30 days and exceptions vary by feature/account. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 7, "href": "/providers/openai-api/#privacy" }, "openness": { "summary": "OpenAI’s SDK and model cards disclose interface and selected safety information, but the API and model weights remain closed and cannot be self-hosted.", "sourceUrls": [ "https://github.com/openai/openai-python", "https://deploymentsafety.openai.com/" ], "confidence": "high", "unknowns": "SDK code and system-card disclosures do not make the hosted model weights, full training data, or complete training process inspectable.", "suggestedScore": 3, "scoreReason": "A public SDK improves API access, but the model and serving service remain closed.", "assessmentType": "conditional", "argument": "A public SDK improves API access, but the model and serving service remain closed. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 3, "href": "/providers/openai-api/#openness" }, "labour": { "summary": "OpenAI’s supplier code sets broad labor standards but does not publish worker outcomes. TIME’s investigation reviewed OpenAI–Sama contracts, payslips, and interviews about 2021–22 Kenyan toxic-content labeling; workers described low take-home pay and distress, while Sama disputed workload/counseling claims and OpenAI said Sama managed pay and support. TIME later reported OpenAI said it was unaware of the conditions and no longer worked with Sama; this is a historical contractor case, not evidence of current API annotation conditions.", "sourceUrls": [ "https://openai.com/policies/supplier-code/", "https://time.com/6247678/openai-chatgpt-kenya-workers/", "https://time.com/7012787/mophat-okinyi/" ], "confidence": "medium", "unknowns": "The sources do not establish whether OpenAI uses other current annotation vendors, present worker outcomes, or the effectiveness of the supplier code across vendors.", "suggestedScore": 4, "scoreReason": "A documented past supplier episode includes worker reports of low pay and psychological distress, countered by Sama and OpenAI responses; the current supplier code has no public independent outcome audit.", "assessmentType": "evidence", "argument": "The supplier code is a useful standard, but the score also accounts for TIME’s 2021–22 reporting based on contracts, payslips, and worker interviews that described low pay and distress; Sama disputed several details, and OpenAI said it was unaware and no longer used Sama. Treat this as a historical supplier concern rather than proof of present conditions, and require current independent vendor-level outcomes.", "score": 4, "href": "/providers/openai-api/#labour" }, "military-contracts": { "summary": "OpenAI publicly describes a 2026 classified Department of War agreement with cloud-only deployment, retained safety controls, cleared staff, and contractual limits. It separately says ChatGPT will reach GenAI.mil for unclassified work used by 3 million civilian and military personnel. Both are direct defense deployments; limits are company-reported and not independently audited.", "sourceUrls": [ "https://openai.com/index/our-agreement-with-the-department-of-war/", "https://openai.com/index/bringing-chatgpt-to-genaimil/" ], "confidence": "high", "unknowns": "Actual use, oversight effectiveness, and compliance with contract limits are not independently established.", "suggestedScore": 5, "scoreReason": "OpenAI reports cloud-only deployment, retained safety controls, cleared staff, and contractual limits; these are meaningful safeguards, but no independent check of enforcement is public.", "assessmentType": "evidence", "argument": "OpenAI reports cloud-only deployment, retained safety controls, cleared staff, and contractual limits; these are meaningful safeguards, but no independent check of enforcement is public. Request the named contract and actual model/endpoint scope with enforceable use controls; do not treat ordinary government work or silence as proof of military use.", "score": 5, "href": "/providers/openai-api/#military-contracts" }, "israel-palestine": { "summary": "We did not find OpenAI, L.L.C. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "OpenAI, L.L.C. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. OpenAI’s April 2026 agreement names Microsoft as its primary cloud partner and a major shareholder, with revenue-share payments to Microsoft through 2030. That is a documented commercial link to a BDS target, not a separate official listing of OpenAI; local use of downloaded weights does not by itself establish a new payment to either company.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://openai.com/index/next-phase-of-microsoft-partnership/" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": { "label": "Microsoft partnership", "sourceIds": [ "openai-ms" ], "argument": "OpenAI’s April 2026 agreement names Microsoft as its primary cloud partner and a major shareholder, with revenue-share payments to Microsoft through 2030. That is a documented commercial link to a BDS target, not a separate official listing of OpenAI; local use of downloaded weights does not by itself establish a new payment to either company." }, "businessContext": null, "score": null, "href": "/providers/openai-api/#israel-palestine" } }, "sources": [ { "url": "https://developers.openai.com/api/docs/guides/your-data", "title": "Data controls in the OpenAI platform", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/our-agreement-with-the-department-of-war/", "title": "Our agreement with the Department of War", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/policies/supplier-code/", "title": "OpenAI Supplier Code of Conduct", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/openai-joins-ports-pike-project/", "title": "OpenAI joins PORTS-Pike project", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/bringing-chatgpt-to-genaimil/", "title": "Bringing ChatGPT to GenAI.mil", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/openai/openai-python", "title": "OpenAI Python API library", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://deploymentsafety.openai.com/", "title": "OpenAI Deployment Safety Hub: System cards", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://time.com/6247678/openai-chatgpt-kenya-workers/", "title": "OpenAI used Kenyan workers on less than $2 per hour: TIME investigation", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://time.com/7012787/mophat-okinyi/", "title": "Mophat Okinyi: TIME100 AI 2024", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://openai.com/index/next-phase-of-microsoft-partnership/", "title": "OpenAI: amended Microsoft partnership, April 27 2026", "kind": "primary", "checkedAt": "2026-10-23" } ], "href": "/providers/openai-api/", "overall": { "score": 4.6, "rated": 5, "total": 6, "coverage": 0.8333333333333334, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 16 } }, { "id": "anthropic-api", "name": "Anthropic API", "category": "providers", "organization": "Anthropic, PBC", "website": "https://www.anthropic.com/api", "summary": "Direct API access to Claude models. Anthropic’s commercial data terms govern the API; model-provider relationship and underlying training corpus remain separate questions.", "scope": "Anthropic develops Claude and operates its API; corporate supplier/compute partnerships are included only where directly disclosed.", "dimensions": { "energy": { "summary": "Anthropic reports a compute agreement with Amazon for up to 5 GW supporting Claude training and deployment. This identifies infrastructure scale/partnering, not actual energy mix, water, emissions per request, or a specific API serving path.", "sourceUrls": [ "https://www.anthropic.com/news/anthropic-amazon-compute" ], "confidence": "medium", "unknowns": "No model-, region-, or request-level footprint established.", "suggestedScore": 5, "scoreReason": "The company discloses very large planned compute commitments, but no current energy mix or request-level footprint; this is a limited corporate-scale claim.", "assessmentType": "conditional", "argument": "The company discloses very large planned compute commitments, but no current energy mix or request-level footprint; this is a limited corporate-scale claim. For a deployment, request or export the selected service and region’s usage-linked footprint when available; do not treat company-wide renewable matching as per-request impact.", "score": 5, "href": "/providers/anthropic-api/#energy" }, "training-rights": { "summary": "Anthropic’s legal center lists separate consumer and commercial terms; its privacy-center disclosure says commercial chats/coding sessions are not used for training absent opt-in/feedback. These customer-data rules do not establish provenance or permissions for Claude’s pretraining corpus.", "sourceUrls": [ "https://www.anthropic.com/legal", "https://privacy.anthropic.com/en/articles/7996885-how-do-you-use-personal-data-in-model-training" ], "confidence": "medium", "unknowns": "Dataset-by-dataset provenance, consent, licensing, compensation, and dispute outcomes remain outside these API data terms.", "suggestedScore": null, "scoreReason": "Commercial customer-data restrictions do not disclose the rights or provenance of Claude’s training corpus; no corpus-rights due-diligence commitment was established.", "assessmentType": "conditional", "argument": "Anthropic’s legal center lists separate consumer and commercial terms; its privacy-center disclosure says commercial chats/coding sessions are not used for training absent opt-in/feedback. These customer-data rules do not establish provenance or permissions for Claude’s pretraining corpus. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/anthropic-api/#training-rights" }, "privacy": { "summary": "Anthropic states commercial offering chats and coding sessions are not used to train models absent explicit opt-in or specified feedback; commercial terms and retention controls are separate from consumer offerings.", "sourceUrls": [ "https://www.anthropic.com/legal", "https://privacy.anthropic.com/en/articles/7996885-how-do-you-use-personal-data-in-model-training" ], "confidence": "medium", "unknowns": "Retention periods and features vary by API/contract, eligible ZDR settings, and reported feedback.", "suggestedScore": 7, "scoreReason": "Commercial data is not used for training absent opt-in/feedback and retention rules are described, but consumer/commercial terms and feature settings differ.", "assessmentType": "conditional", "argument": "Commercial data is not used for training absent opt-in/feedback and retention rules are described, but consumer/commercial terms and feature settings differ. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 7, "href": "/providers/anthropic-api/#privacy" }, "openness": { "summary": "Anthropic publishes SDKs and model cards, but Claude is a closed hosted service with no public weights or self-hosting path in the reviewed sources.", "sourceUrls": [ "https://github.com/anthropics/anthropic-sdk-python", "https://www.anthropic.com/system-cards" ], "confidence": "high", "unknowns": "System cards are selective disclosures; model weights, full training data, and complete training methods remain unavailable in the reviewed sources.", "suggestedScore": 3, "scoreReason": "SDK and system-card disclosures do not make Claude weights or serving infrastructure portable.", "assessmentType": "conditional", "argument": "SDK and system-card disclosures do not make Claude weights or serving infrastructure portable. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 3, "href": "/providers/anthropic-api/#openness" }, "labour": { "summary": "Anthropic’s 2025 Claude 4 system card and inbound-services agreement disclose use of paid data-labeling contractors and require crowd-work vendors to follow minimum worker-wellbeing expectations. Stanford FMTI found no public worker compensation, location, or top-source labor-practice outcomes, so the standards remain policy commitments rather than independently verified results.", "sourceUrls": [ "https://www-cdn.anthropic.com/de8ba9b01c9ab7cbabf5c33b80b7bbc618857627/Model%5C_Card%5C_Claude%5C_3.pdf", "https://crfm.stanford.edu/fmti/December-2025/company-reports/Anthropic_FinalReport_FMTI2025.html", "https://www.anthropic.com/legal/inbound-services-agreement" ], "confidence": "medium", "unknowns": "Current pay, vendor locations, worker exposure, and compliance/audit results are not published per major data source.", "suggestedScore": 4, "scoreReason": "Public vendor-wellbeing standards and contractor disclosure are meaningful, but Stanford FMTI found no compensation or worker-level outcomes for the principal data sources.", "assessmentType": "conditional", "argument": "Anthropic discloses contractor data work and vendor wellbeing requirements, while Stanford FMTI found no public pay, location, or worker-outcome data for leading sources. The commitments are relevant but not independently outcome-checked; request current vendor-level evidence before treating them as worker protections.", "score": 4, "href": "/providers/anthropic-api/#labour" }, "military-contracts": { "summary": "Anthropic reports a two-year DoD CDAO prototype agreement with a $200 million ceiling, Claude in classified Palantir mission workflows, and Claude Gov deployments on AWS for national-security agencies. In March 2026, Anthropic said DoW designated it a supply-chain risk, narrowly applying to Claude use directly under DoW contracts, and that it would support a transition for warfighters.", "sourceUrls": [ "https://www.anthropic.com/news/anthropic-and-the-department-of-defense-to-advance-responsible-ai-in-defense-operations", "https://www.anthropic.com/news/where-stand-department-war" ], "confidence": "high", "unknowns": "These company statements do not establish the specific operational outcomes or independently assess the safeguards of deployments.", "suggestedScore": 3, "scoreReason": "The company documents classified defense use and a $200 million prototype agreement; deployments and guardrails are not independently assessed.", "assessmentType": "evidence", "argument": "The company documents classified defense use and a $200 million prototype agreement; deployments and guardrails are not independently assessed. Request the named contract and actual model/endpoint scope with enforceable use controls; do not treat ordinary government work or silence as proof of military use.", "score": 3, "href": "/providers/anthropic-api/#military-contracts" }, "israel-palestine": { "summary": "We did not find Anthropic, PBC among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Anthropic, PBC is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Anthropic names AWS as its primary training and cloud provider and reports a large compute commitment and Amazon investment. This connects its business to a BDS pressure target without making Anthropic a separately named target; the announcement does not identify the host of every Claude request.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://www.anthropic.com/news/anthropic-amazon-compute" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": { "label": "Amazon partnership", "sourceIds": [ "anthropic-aws" ], "argument": "Anthropic names AWS as its primary training and cloud provider and reports a large compute commitment and Amazon investment. This connects its business to a BDS pressure target without making Anthropic a separately named target; the announcement does not identify the host of every Claude request." }, "businessContext": null, "score": null, "href": "/providers/anthropic-api/#israel-palestine" } }, "sources": [ { "url": "https://www.anthropic.com/legal", "title": "Anthropic Legal Center", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://privacy.anthropic.com/en/articles/7996885-how-do-you-use-personal-data-in-model-training", "title": "How Anthropic uses personal data in model training", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/news/anthropic-amazon-compute", "title": "Anthropic and Amazon expand compute collaboration", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www-cdn.anthropic.com/de8ba9b01c9ab7cbabf5c33b80b7bbc618857627/Model%5C_Card%5C_Claude%5C_3.pdf", "title": "Claude 3 Model Card", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/news/anthropic-and-the-department-of-defense-to-advance-responsible-ai-in-defense-operations", "title": "Anthropic and the Department of Defense to advance responsible AI in defense operations", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/news/where-stand-department-war", "title": "Where things stand with the Department of War", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://github.com/anthropics/anthropic-sdk-python", "title": "Anthropic Python SDK", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/system-cards", "title": "Anthropic system cards", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://crfm.stanford.edu/fmti/December-2025/company-reports/Anthropic_FinalReport_FMTI2025.html", "title": "Stanford FMTI 2025 Anthropic transparency assessment", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://www.anthropic.com/legal/inbound-services-agreement", "title": "Anthropic Inbound Services Agreement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/anthropic-api/", "overall": { "score": 4.4, "rated": 5, "total": 6, "coverage": 0.8333333333333334, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 17 } }, { "id": "google-vertex-ai", "name": "Google Vertex AI", "category": "providers", "organization": "Google LLC / Google Cloud", "website": "https://cloud.google.com/vertex-ai", "summary": "Managed Google Cloud platform hosting Google and third-party models. Google Cloud privacy and infrastructure commitments do not necessarily cover third-party model corpus rights.", "scope": "Google is cloud host and also model provider for its own models; third-party models available in Vertex have model-specific terms.", "dimensions": { "energy": { "summary": "Google Cloud Carbon Footprint now includes Vertex AI and allocates customer emissions by product, project, and region at monthly granularity; its 2026 release notes describe model-inference emissions being assigned to cloud services, especially Vertex AI. This is more actionable than a corporate renewable claim, but the allocation is estimated, not third-party assured, and is not a per-request measurement.", "sourceUrls": [ "https://cloud.google.com/infrastructure", "https://docs.cloud.google.com/carbon-footprint/docs/methodology", "https://docs.cloud.google.com/carbon-footprint/docs/view-carbon-data?hl=en", "https://docs.cloud.google.com/carbon-footprint/docs/release-notes" ], "confidence": "medium", "unknowns": "The customer report is monthly and allocation-based; water and per-request usage are not reported as direct model metrics, some grid data rely on annual country averages, and Google says its customer-specific report is not third-party assured.", "suggestedScore": 6, "scoreReason": "Customer- and region-specific estimates cover Vertex AI and include inference allocation, but remain monthly, method-allocated estimates without third-party assurance or per-request energy/water values.", "assessmentType": "conditional", "argument": "Customer- and region-specific estimates cover Vertex AI and include inference allocation, but remain monthly, method-allocated estimates without third-party assurance or per-request energy/water values. For a deployment, request or export the selected service and region’s usage-linked footprint when available; do not treat company-wide renewable matching as per-request impact.", "score": 6, "href": "/providers/google-vertex-ai/#energy" }, "training-rights": { "summary": "Vertex AI terms state Google will not use customer data to train or fine-tune AI/ML models without prior permission or instruction. That is a hosting/data-use commitment and does not establish training-data rights for Google or third-party models.", "sourceUrls": [ "https://docs.cloud.google.com/vertex-ai/generative-ai/docs/vertex-ai-zero-data-retention" ], "confidence": "high", "unknowns": "Training-corpus provenance/consent remains model-specific and not resolved by the host commitment.", "suggestedScore": null, "scoreReason": "Vertex’s customer-input restriction is about hosting and customer data, not rights in Google or third-party model training corpora.", "assessmentType": "conditional", "argument": "Vertex AI terms state Google will not use customer data to train or fine-tune AI/ML models without prior permission or instruction. That is a hosting/data-use commitment and does not establish training-data rights for Google or third-party models. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/google-vertex-ai/#training-rights" }, "privacy": { "summary": "Google’s Vertex AI documentation describes training restriction and privacy commitments; zero-retention eligibility is feature-specific and managed-model behavior may differ for safety or stateful features.", "sourceUrls": [ "https://docs.cloud.google.com/vertex-ai/generative-ai/docs/vertex-ai-zero-data-retention" ], "confidence": "medium", "unknowns": "Exact endpoint/model, enabled feature, region, and retention configuration need verification.", "suggestedScore": 7, "scoreReason": "Google states customer data is not used to train without permission and offers zero-retention paths, but eligibility and stateful/safety feature behavior vary.", "assessmentType": "conditional", "argument": "Google states customer data is not used to train without permission and offers zero-retention paths, but eligibility and stateful/safety feature behavior vary. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 7, "href": "/providers/google-vertex-ai/#privacy" }, "openness": { "summary": "Vertex offers some open models and deployment choices, but the service stack is managed and Gemini remains closed; portability depends on the model and license.", "sourceUrls": [ "https://cloud.google.com/vertex-ai/generative-ai/docs/model-garden/explore-models" ], "confidence": "high", "unknowns": "Access to an open model in a managed catalog does not make all Vertex models open or establish training-data provenance.", "suggestedScore": 5, "scoreReason": "Open-model choices and some deployment control help; managed infrastructure and closed first-party models limit portability.", "assessmentType": "conditional", "argument": "Open-model choices and some deployment control help; managed infrastructure and closed first-party models limit portability. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 5, "href": "/providers/google-vertex-ai/#openness" }, "labour": { "summary": "Google’s supplier code applies labor and human-rights requirements across direct, temporary, migrant, contract, and subcontracted workers; it allows audits, requires remediation, and sets a 60-hour weekly limit with stated exceptions. This is a supplier policy, not an audit of AI data workers.", "sourceUrls": [ "https://about.google/company-info/supplier-code-of-conduct/" ], "confidence": "high", "unknowns": "No current AI-annotation-specific outcomes, public audit findings, or remediation data for Vertex-related data work were established.", "suggestedScore": 4, "scoreReason": "Google publishes broad supplier protections and audit rights, but this review found no worker-level results for AI data work.", "assessmentType": "evidence", "argument": "Google publishes broad supplier protections and audit rights, but this review found no worker-level results for AI data work. Ask for current vendor-level pay, exposure, location, audit, and remediation records; a supplier code cannot substitute for worker outcomes.", "score": 4, "href": "/providers/google-vertex-ai/#labour" }, "military-contracts": { "summary": "Google documents access for Department of War personnel and classified defense workflows. Separately, Israel’s procurement administration identifies Google as a Nimbus public-cloud provider; AFSC and Who Profits describe military/security use, while the Guardian reports contract terms limiting service withdrawal. These sources do not establish Google’s role in a particular attack.", "sourceUrls": [ "https://cloud.google.com/blog/topics/public-sector/announcements-and-highlights-from-google-public-sector-summit-2026", "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/alphabet" ], "confidence": "high", "unknowns": "The source does not disclose contract terms or evaluate downstream deployment outcomes.", "suggestedScore": 3, "scoreReason": "Google documents U.S. Department of War and classified defense workflows; Israel’s procurement record identifies Google as a Nimbus cloud provider, and independent research/reporting describes military or security use and limits on withdrawal. The reviewed evidence does not attribute a specific harmful operation to Google.", "assessmentType": "conditional", "argument": "Google documents U.S. Department of War and classified workflows. Israel’s procurement record identifies Google as a Project Nimbus provider; AFSC and Who Profits describe military/security use, while the Guardian reports contract terms limiting service withdrawal. These records establish relationships and reported use, not Google’s role in a particular operation; request the applicable contract, endpoint scope, and enforceable controls.", "score": 3, "href": "/providers/google-vertex-ai/#military-contracts" }, "israel-palestine": { "summary": "Google / Alphabet is named as a pressure target in the BNC’s official campaign sources. The product is developed or supplied by Google / Alphabet.", "argument": "If following this BNC campaign is a requirement, consider alternatives to paid Google / Alphabet products where practical. Your selected model, cloud and billing route can introduce other companies into the purchase. This label follows the named company’s campaign status; it does not claim that BDS separately reviewed this exact product.", "assessmentType": "evidence", "confidence": "high", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "pressure", "statusLabel": "Pressure target", "reviewedAt": "2026-10-23", "assessmentScope": "Developer / service company", "bdsMatch": { "target": "Google / Alphabet", "status": "pressure", "scope": "Developer / service company", "relationship": "The product is developed or supplied by Google / Alphabet.", "application": "If following this BNC campaign is a requirement, consider alternatives to paid Google / Alphabet products where practical.", "sourceIds": [] }, "upstreamExposure": null, "businessContext": { "summary": "Israel’s procurement administration says Google was selected with AWS for Nimbus public-cloud services and lists the Ministry of Defense and IDF among project partners. The Guardian’s investigation of leaked agreement documents reports that Nimbus terms limited Google’s ability to suspend service, including after alleged policy violations. Who Profits and AFSC describe cloud and AI use by military/security bodies, including surveillance-related capabilities; these are attributed research claims, not proof of Google’s role in a specific attack. Both companies denied circumventing legal obligations in response to the Guardian investigation.", "unknowns": "The government record establishes the contract and project participants; the investigative reporting describes contract controls, and research organizations make broader use claims. The reviewed evidence does not establish Google’s contribution to a particular operation, injury, or other specific harm.", "sourceUrls": [ "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/alphabet" ] }, "score": null, "href": "/providers/google-vertex-ai/#israel-palestine" } }, "sources": [ { "url": "https://docs.cloud.google.com/vertex-ai/generative-ai/docs/vertex-ai-zero-data-retention", "title": "Vertex AI and zero data retention", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cloud.google.com/infrastructure", "title": "Google Cloud infrastructure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cloud.google.com/blog/topics/inside-google-cloud/google-cloud-selected-to-provide-cloud-services-to-the-state-of-israel", "title": "Google Cloud selected for Project Nimbus", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cloud.google.com/blog/topics/public-sector/announcements-and-highlights-from-google-public-sector-summit-2026", "title": "Announcements and highlights from Google Public Sector Summit 2026", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://about.google/company-info/supplier-code-of-conduct/", "title": "Google Supplier Code of Conduct", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cloud.google.com/vertex-ai/generative-ai/docs/model-garden/explore-models", "title": "Overview of Model Garden", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://mr.gov.il/ilgstorefront/en/news/details/111222", "title": "Nimbus Tender — Israel Government Procurement Administration", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "title": "Revealed: Israel demanded Google and Amazon use secret “wink” to sidestep legal orders", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://www.whoprofits.org/publications/report/164", "title": "Digitizing Occupation: The Role of Big Tech in Israeli Military Infrastructure", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://investigate.afsc.org/company/alphabet", "title": "Alphabet / Google — AFSC Investigate", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://docs.cloud.google.com/carbon-footprint/docs/methodology", "title": "Carbon Footprint reporting methodology", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.cloud.google.com/carbon-footprint/docs/view-carbon-data?hl=en", "title": "View Carbon Footprint data", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.cloud.google.com/carbon-footprint/docs/release-notes", "title": "Carbon Footprint release notes", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/google-vertex-ai/", "overall": { "score": 5, "rated": 5, "total": 6, "coverage": 0.8333333333333334, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 14 } }, { "id": "aws-bedrock", "name": "AWS Bedrock", "category": "providers", "organization": "Amazon Web Services, Inc.", "website": "https://aws.amazon.com/bedrock/", "summary": "Managed model marketplace and inference service; AWS hosting terms differ from each underlying model provider’s training data and IP practices.", "scope": "AWS operates infrastructure, routes selected third-party and Amazon models, and customer settings can affect retention/provider sharing.", "dimensions": { "energy": { "summary": "AWS Sustainability estimates carbon and water impacts for cloud usage and supports account, region, and service groupings; Bedrock is not separately broken out in current service views and falls within “Other.” The former Customer Carbon Footprint Tool was deprecated in June 2026, so the current Sustainability service is the relevant source. These estimates are useful for account-level review, not Bedrock model/request-level electricity or water.", "sourceUrls": [ "https://aws.amazon.com/sustainability/", "https://docs.aws.amazon.com/sustainability/latest/userguide/methodology.html", "https://docs.aws.amazon.com/sustainability/latest/userguide/csv-reports.html", "https://docs.aws.amazon.com/sustainability/latest/APIReference/API_EstimatedCarbonEmissions.html", "https://docs.aws.amazon.com/ccft/latest/releasenotes/what-is-ccftrn.html" ], "confidence": "medium", "unknowns": "No Bedrock request-level power, water or emissions figure; workload and region determine actual footprint.", "suggestedScore": 5, "scoreReason": "AWS now offers independently verified estimation methods and account/region reporting, but Bedrock is aggregated with other products and the figures are not model- or request-specific.", "assessmentType": "conditional", "argument": "AWS now offers independently verified estimation methods and account/region reporting, but Bedrock is aggregated with other products and the figures are not model- or request-specific. For a deployment, request or export the selected service and region’s usage-linked footprint when available; do not treat company-wide renewable matching as per-request impact.", "score": 5, "href": "/providers/aws-bedrock/#energy" }, "training-rights": { "summary": "AWS says Bedrock prompts/outputs are not used to train underlying models absent customer consent, but the training corpus and rights of each third-party foundation model remain provider-specific.", "sourceUrls": [ "https://aws.amazon.com/bedrock/amazon-models/privacy/" ], "confidence": "high", "unknowns": "Review selected model-provider terms and dataset provenance separately; customer prompt policy is not corpus-rights evidence.", "suggestedScore": null, "scoreReason": "Bedrock prompt/output protections do not establish rights in third-party models’ original corpora; no catalog-wide corpus due-diligence evidence was found.", "assessmentType": "conditional", "argument": "AWS says Bedrock prompts/outputs are not used to train underlying models absent customer consent, but the training corpus and rights of each third-party foundation model remain provider-specific. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/aws-bedrock/#training-rights" }, "privacy": { "summary": "Bedrock account/project retention modes include none/default/inherit and provider_data_share; some models or APIs require retention or provider sharing, and flagged safety content can be retained/reviewed. Settings are configuration- and model-specific.", "sourceUrls": [ "https://aws.amazon.com/blogs/security/enforce-zero-data-retention-on-amazon-bedrock-with-bedrock-projects-and-service-control-policies/" ], "confidence": "high", "unknowns": "Model’s current allowed modes, API statefulness, region/cross-region routing, and account configuration must be inspected.", "suggestedScore": 6, "scoreReason": "Retention and provider-sharing controls are configurable, but behavior varies by model/API and safety-review exceptions can retain content.", "assessmentType": "evidence", "argument": "Retention and provider-sharing controls are configurable, but behavior varies by model/API and safety-review exceptions can retain content. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 6, "href": "/providers/aws-bedrock/#privacy" }, "openness": { "summary": "Bedrock offers open-weight and proprietary models through a managed service; model choice is broad, but the serving stack and portability depend on each provider.", "sourceUrls": [ "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "confidence": "high", "unknowns": "Open-weight availability and rights vary by provider/model and do not carry across the whole service.", "suggestedScore": 5, "scoreReason": "Open-weight choice is useful, but the managed inference stack and model-specific terms limit portability.", "assessmentType": "conditional", "argument": "Open-weight choice is useful, but the managed inference stack and model-specific terms limit portability. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 5, "href": "/providers/aws-bedrock/#openness" }, "labour": { "summary": "Amazon’s Supply Chain Standards Manual makes human-rights and labor standards a condition of supplier relationships and describes monitoring and corrective processes. This is a supply-chain policy, not evidence of current Bedrock annotation-worker conditions or audit outcomes.", "sourceUrls": [ "https://sustainability.aboutamazon.com/amazon_supplier_manual_english.pdf" ], "confidence": "high", "unknowns": "No current Bedrock- or foundation-model-specific labor audit results were established.", "suggestedScore": 4, "scoreReason": "Amazon requires supplier standards and describes monitoring, but publishes no Bedrock-specific worker outcomes here.", "assessmentType": "evidence", "argument": "Amazon requires supplier standards and describes monitoring, but publishes no Bedrock-specific worker outcomes here. Ask for current vendor-level pay, exposure, location, audit, and remediation records; a supplier code cannot substitute for worker outcomes.", "score": 4, "href": "/providers/aws-bedrock/#labour" }, "military-contracts": { "summary": "AWS funds U.S. defense work that names autonomous systems and battle-management uses. Separately, Israel’s procurement administration identifies AWS as a Nimbus public-cloud provider; AFSC and Who Profits describe military/security use, while the Guardian reports contract terms limiting service withdrawal. These sources do not establish AWS’s role in a particular attack.", "sourceUrls": [ "https://aws.amazon.com/blogs/publicsector/aws-announces-up-to-100-million-in-federal-credits-to-accelerate-innovation-for-national-security-and-scientific-missions/", "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/amazon" ], "confidence": "high", "unknowns": "The announced credits and eligible use cases do not establish which projects received funding or which AWS AI services each project selected.", "suggestedScore": 2, "scoreReason": "AWS funds defense work that explicitly includes autonomous systems and battle management across classification levels; no independent safeguards assessment is published.", "assessmentType": "conditional", "argument": "AWS funds defense work that explicitly includes autonomous systems and battle management across classification levels; no independent safeguards assessment is published. Request the named contract and actual model/endpoint scope with enforceable use controls; do not treat ordinary government work or silence as proof of military use.", "score": 2, "href": "/providers/aws-bedrock/#military-contracts" }, "israel-palestine": { "summary": "Amazon is named as a pressure target in the BNC’s official campaign sources. The product is developed or supplied by Amazon.", "argument": "If following this BNC campaign is a requirement, consider alternatives to paid Amazon products where practical. Your selected model, cloud and billing route can introduce other companies into the purchase. This label follows the named company’s campaign status; it does not claim that BDS separately reviewed this exact product.", "assessmentType": "evidence", "confidence": "high", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "pressure", "statusLabel": "Pressure target", "reviewedAt": "2026-10-23", "assessmentScope": "Developer / service company", "bdsMatch": { "target": "Amazon", "status": "pressure", "scope": "Developer / service company", "relationship": "The product is developed or supplied by Amazon.", "application": "If following this BNC campaign is a requirement, consider alternatives to paid Amazon products where practical.", "sourceIds": [] }, "upstreamExposure": null, "businessContext": { "summary": "Israel’s procurement administration names AWS as a Nimbus public-cloud provider and the Ministry of Defense and IDF as project partners. The Guardian’s investigation of leaked agreement documents reports terms restricting AWS’s ability to suspend service, including after alleged policy violations. Who Profits and AFSC describe AWS cloud services in military/security infrastructure and surveillance-related uses; those are attributed research claims, not proof of AWS’s role in a specific attack. Both companies denied circumventing legal obligations in response to the Guardian investigation.", "unknowns": "The government record establishes the contract and project participants; the investigative reporting describes contract controls, and research organizations make broader use claims. The reviewed evidence does not establish AWS’s contribution to a particular operation, injury, or other specific harm.", "sourceUrls": [ "https://mr.gov.il/ilgstorefront/en/news/details/111222", "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "https://www.whoprofits.org/publications/report/164", "https://investigate.afsc.org/company/amazon" ] }, "score": null, "href": "/providers/aws-bedrock/#israel-palestine" } }, "sources": [ { "url": "https://aws.amazon.com/bedrock/amazon-models/privacy/", "title": "Amazon Model Training & Privacy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://aws.amazon.com/blogs/security/enforce-zero-data-retention-on-amazon-bedrock-with-bedrock-projects-and-service-control-policies/", "title": "Enforce zero data retention on Amazon Bedrock", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://aws.amazon.com/sustainability/", "title": "AWS Sustainability", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://aws.amazon.com/about-aws/whats-new/2025/09/amazon-bedrock-israel-tel-aviv-region/", "title": "Amazon Bedrock available in Israel (Tel Aviv) Region", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://press.aboutamazon.com/2021/6/aws-to-open-data-centers-in-israel", "title": "AWS to Open Data Centers in Israel", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://sustainability.aboutamazon.com/aws-sustainability-fact-sheets/aws-fact-sheet-israel.pdf", "title": "AWS Israel Sustainability Fact Sheet", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://aws.amazon.com/blogs/publicsector/aws-announces-up-to-100-million-in-federal-credits-to-accelerate-innovation-for-national-security-and-scientific-missions/", "title": "AWS announces up to $100 million in federal credits for national security and scientific missions", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://sustainability.aboutamazon.com/amazon_supplier_manual_english.pdf", "title": "Amazon Supply Chain Standards Manual", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html", "title": "Amazon Bedrock model cards", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://mr.gov.il/ilgstorefront/en/news/details/111222", "title": "Nimbus Tender — Israel Government Procurement Administration", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.theguardian.com/us-news/2025/oct/29/google-amazon-israel-contract-secret-code", "title": "Revealed: Israel demanded Google and Amazon use secret “wink” to sidestep legal orders", "kind": "reporting", "checkedAt": "2026-10-23" }, { "url": "https://www.whoprofits.org/publications/report/164", "title": "Digitizing Occupation: The Role of Big Tech in Israeli Military Infrastructure", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://investigate.afsc.org/company/amazon", "title": "Amazon — AFSC Investigate", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://docs.aws.amazon.com/sustainability/latest/userguide/methodology.html", "title": "AWS Sustainability calculation methodology", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.aws.amazon.com/sustainability/latest/userguide/csv-reports.html", "title": "AWS Sustainability CSV reports", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.aws.amazon.com/sustainability/latest/APIReference/API_EstimatedCarbonEmissions.html", "title": "Estimated carbon emissions API", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.aws.amazon.com/ccft/latest/releasenotes/what-is-ccftrn.html", "title": "Customer Carbon Footprint Tool release notes", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/aws-bedrock/", "overall": { "score": 4.4, "rated": 5, "total": 6, "coverage": 0.8333333333333334, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 18 } }, { "id": "azure-ai-foundry", "name": "Azure AI Foundry", "category": "providers", "organization": "Microsoft Corporation", "website": "https://azure.microsoft.com/products/ai-foundry", "summary": "Microsoft-hosted models and partner model catalog in Microsoft Foundry. Scope differs across Azure-operated vs partner-operated offerings.", "scope": "Record covers Models sold by Azure; third-party partner offerings may have distinct data, model rights, processing, and contractual terms.", "dimensions": { "energy": { "summary": "Azure Carbon Optimization reports monthly emissions for all Azure resource types based on billing and usage, with API access and 12 months of history. That gives Foundry buyers a route to inspect the Azure resource footprint, but the estimate is neither a model-level nor per-request power/water measurement and requires resource/account attribution.", "sourceUrls": [ "https://learn.microsoft.com/en-us/azure/carbon-optimization/overview" ], "confidence": "medium", "unknowns": "Monthly Azure emissions are based on billing and usage and can be reviewed at resource granularity, but do not isolate individual model inference, token counts, or water consumed by a request.", "suggestedScore": 6, "scoreReason": "A customer can examine resource-level Azure emissions across all resource types, but reporting is monthly and does not isolate Foundry models or individual inference requests.", "assessmentType": "conditional", "argument": "A customer can examine resource-level Azure emissions across all resource types, but reporting is monthly and does not isolate Foundry models or individual inference requests. For a deployment, request or export the selected service and region’s usage-linked footprint when available; do not treat company-wide renewable matching as per-request impact.", "score": 6, "href": "/providers/azure-ai-foundry/#energy" }, "training-rights": { "summary": "For Azure-hosted Models sold by Azure, Microsoft says prompts, outputs and customer training data are not used by model providers or Microsoft to train foundation models without permission/instruction. This is a hosting data-use commitment, not evidence about training corpus licenses.", "sourceUrls": [ "https://learn.microsoft.com/en-us/azure/ai-foundry/responsible-ai/openai/data-privacy" ], "confidence": "high", "unknowns": "Partner models and corpus provenance are outside this specific commitment and must be reviewed separately.", "suggestedScore": null, "scoreReason": "Microsoft’s hosting promise limits customer-data training, not the provenance or licensing of hosted model corpora.", "assessmentType": "conditional", "argument": "For Azure-hosted Models sold by Azure, Microsoft says prompts, outputs and customer training data are not used by model providers or Microsoft to train foundation models without permission/instruction. This is a hosting data-use commitment, not evidence about training corpus licenses. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/azure-ai-foundry/#training-rights" }, "privacy": { "summary": "Microsoft states Azure-operated models process data in the customer-selected geography for standard deployments, while Global/DataZone deployments can process across a wider area; stateful features and abuse monitoring can store data.", "sourceUrls": [ "https://learn.microsoft.com/en-us/azure/ai-foundry/responsible-ai/openai/data-privacy" ], "confidence": "high", "unknowns": "Actual deployment type, geography, feature and abuse-monitoring eligibility/configuration require confirmation.", "suggestedScore": 7, "scoreReason": "Azure documents customer-selected geography for standard deployment and strong training restrictions, with broader processing in global/data-zone modes and storage in some stateful/abuse-monitoring cases.", "assessmentType": "evidence", "argument": "Azure documents customer-selected geography for standard deployment and strong training restrictions, with broader processing in global/data-zone modes and storage in some stateful/abuse-monitoring cases. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 7, "href": "/providers/azure-ai-foundry/#privacy" }, "openness": { "summary": "Foundry includes some models that can be deployed on managed compute, alongside proprietary hosted offerings; code, weights, and portability vary by model.", "sourceUrls": [ "https://learn.microsoft.com/en-us/azure/foundry/concepts/foundry-models-overview", "https://learn.microsoft.com/en-us/azure/foundry/responsible-ai/openai/transparency-note" ], "confidence": "high", "unknowns": "Catalog size and model cards do not establish open weights or data provenance for each entry; licenses and deployment control vary.", "suggestedScore": 5, "scoreReason": "Some managed-compute deployments give customers more control, but the catalog also contains closed services and licenses vary.", "assessmentType": "evidence", "argument": "Some managed-compute deployments give customers more control, but the catalog also contains closed services and licenses vary. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 5, "href": "/providers/azure-ai-foundry/#openness" }, "labour": { "summary": "Microsoft’s FY2025 Human Rights Transparency Report describes supplier monitoring and reports findings in its Devices/Cloud Supply Chain and Provisioning scope, including 34 forced-labor/trafficking indicators (33 closed, one corrective plan), 50 fair-compensation findings (46 closed, four plans), and 75 working-hours/rest findings (68 closed, seven plans). These are supplier-audit findings, not AI-annotation-specific results.", "sourceUrls": [ "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Human-Rights-Transparency-Report-2025.pdf" ], "confidence": "high", "unknowns": "The published findings cover specified supplier audit scopes, not all suppliers, model evaluation labor, or all remediation outcomes.", "suggestedScore": 6, "scoreReason": "Microsoft reports supplier findings and closure plans, but the reported scope is not specific to AI annotation and some plans remain open.", "assessmentType": "evidence", "argument": "Microsoft reports supplier findings and closure plans, but the reported scope is not specific to AI annotation and some plans remain open. Ask for current vendor-level pay, exposure, location, audit, and remediation records; a supplier code cannot substitute for worker outcomes.", "score": 6, "href": "/providers/azure-ai-foundry/#labour" }, "military-contracts": { "summary": "Microsoft’s final 2026 review says its external investigation found evidence supporting elements of 2025 reporting, including IMOD Azure storage consumption in the Netherlands and AI-service use. Microsoft says it disabled specified storage/AI subscriptions in September 2025, while continuing other software, professional services, Azure and Azure AI services to IMOD.", "sourceUrls": [ "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Summary-of-2025-External-Investigation-Follow-Up.pdf" ], "confidence": "high", "unknowns": "The report does not access IMOD customer content, does not establish the content/use of every service, and does not say the wider commercial relationship ended.", "suggestedScore": 2, "scoreReason": "The company confirms defense-ministry Azure/AI use and a review supporting elements of serious surveillance-related reporting; it stopped specified services but continues others.", "assessmentType": "evidence", "argument": "The company confirms defense-ministry Azure/AI use and a review supporting elements of serious surveillance-related reporting; it stopped specified services but continues others. Request the named contract and actual model/endpoint scope with enforceable use controls; do not treat ordinary government work or silence as proof of military use.", "score": 2, "href": "/providers/azure-ai-foundry/#military-contracts" }, "israel-palestine": { "summary": "Microsoft is named as a consumer boycott priority target in the BNC guide. The product is developed or supplied by Microsoft.", "argument": "If following this BNC campaign is a requirement, consider alternatives to paid Microsoft products where practical. Your selected model, cloud and billing route can introduce other companies into the purchase. This label follows the named company’s campaign status; it does not claim that BDS separately reviewed this exact product.", "assessmentType": "evidence", "confidence": "high", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://bdsmovement.net/microsoft" ], "bdsStatus": "priority-boycott", "statusLabel": "Boycott target", "reviewedAt": "2026-10-23", "assessmentScope": "Developer / service company", "bdsMatch": { "target": "Microsoft", "status": "priority-boycott", "scope": "Developer / service company", "relationship": "The product is developed or supplied by Microsoft.", "application": "If following this BNC campaign is a requirement, consider alternatives to paid Microsoft products where practical.", "sourceIds": [ "microsoft" ] }, "upstreamExposure": null, "businessContext": { "summary": "Microsoft’s final 2026 review says Covington found evidence supporting elements of reporting about IMOD Azure storage and AI-service use, including Azure storage consumption in the Netherlands. Microsoft says it disabled specified subscriptions/services in September 2025 but continues providing IMOD with other Azure, Azure AI, software, and professional services. The report did not access customer content and does not establish all downstream uses.", "unknowns": "Microsoft describes only selected service suspensions; the report does not establish all IMOD use, civilian impact, or the full scope of continuing services.", "sourceUrls": [ "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Summary-of-2025-External-Investigation-Follow-Up.pdf" ] }, "score": null, "href": "/providers/azure-ai-foundry/#israel-palestine" } }, "sources": [ { "url": "https://learn.microsoft.com/en-us/azure/ai-foundry/responsible-ai/openai/data-privacy", "title": "Data, privacy, and security for Foundry Models sold by Azure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://learn.microsoft.com/en-us/azure/foundry/responsible-ai/openai/transparency-note", "title": "Azure OpenAI Transparency Note", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Summary-of-2025-External-Investigation-Follow-Up.pdf", "title": "Microsoft Summary of 2025 External Investigation and Follow Up", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Human-Rights-Transparency-Report-2025.pdf", "title": "Microsoft Human Rights Transparency Report 2025", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://learn.microsoft.com/en-us/azure/foundry/concepts/foundry-models-overview", "title": "Microsoft Foundry Models overview", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://learn.microsoft.com/en-us/azure/carbon-optimization/overview", "title": "What is Carbon optimization in Azure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/microsoft", "title": "BNC: Microsoft campaign and subsidiaries guidance", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/azure-ai-foundry/", "overall": { "score": 5.2, "rated": 5, "total": 6, "coverage": 0.8333333333333334, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 13 } }, { "id": "together-ai", "name": "Together AI", "category": "providers", "organization": "Together Computer, Inc.", "website": "https://www.together.ai/", "summary": "Independent inference, fine-tuning, and training platform offering open and proprietary model hosting. Model and hosting rights vary by selected model.", "scope": "Provider operates API/infrastructure and can host models by others; customer-data privacy statements do not establish corpus rights.", "dimensions": { "energy": { "summary": "The reviewed sources give no service-level power, emissions, or water figures.", "sourceUrls": [ "https://www.together.ai/blog/together-ai-expands-in-europe" ], "confidence": "medium", "unknowns": "Request-level electricity, emissions, and water use remain unknown.", "suggestedScore": null, "scoreReason": "No verified facility or per-request energy evidence; planned capacity and third-party sustainability claims are insufficient for a rating.", "assessmentType": "conditional", "argument": "The reviewed sources give no service-level power, emissions, or water figures. For a deployment, request or export the selected service and region’s usage-linked footprint when available; do not treat company-wide renewable matching as per-request impact.", "score": null, "href": "/providers/together-ai/#energy" }, "training-rights": { "summary": "Customer-data policy covers service inputs, not legal provenance of the hosted model’s pretraining corpus. Together’s catalog includes third-party/open models whose licenses and provenance differ.", "sourceUrls": [ "https://www.together.ai/privacy" ], "confidence": "low", "unknowns": "Per-model license, training source/consent/compensation, and rights claims need model-specific review.", "suggestedScore": null, "scoreReason": "Prompt-data policy does not settle upstream model-corpus rights, which depend on each hosted model; no across-catalog due-diligence evidence found.", "assessmentType": "argument", "argument": "Customer-data policy covers service inputs, not legal provenance of the hosted model’s pretraining corpus. Together’s catalog includes third-party/open models whose licenses and provenance differ. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/together-ai/#training-rights" }, "privacy": { "summary": "Together’s privacy policy says customer data is not used to train without explicit opt-in and describes settings to avoid retaining prompts, responses or training data.", "sourceUrls": [ "https://www.together.ai/privacy" ], "confidence": "medium", "unknowns": "Verify current product, account, retention settings, and separate terms for fine-tuning/custom model services.", "suggestedScore": 7, "scoreReason": "The company says customer data is not used for training without opt-in and documents no-retention settings, but public verification and feature-level exceptions remain limited.", "assessmentType": "conditional", "argument": "The company says customer data is not used for training without opt-in and documents no-retention settings, but public verification and feature-level exceptions remain limited. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 7, "href": "/providers/together-ai/#privacy" }, "openness": { "summary": "Together serves many open-weight models, but its inference stack remains managed; dedicated deployments do not make the service code portable.", "sourceUrls": [ "https://www.together.ai/models" ], "confidence": "high", "unknowns": "Catalog availability does not establish that each model is OSI-open, that training corpora are documented, or that all service components are open.", "suggestedScore": 5, "scoreReason": "Open weights are available upstream, while Together’s own serving stack remains managed.", "assessmentType": "conditional", "argument": "Open weights are available upstream, while Together’s own serving stack remains managed. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 5, "href": "/providers/together-ai/#openness" }, "labour": { "summary": "Together's careers page describes employee roles and benefits, while its infrastructure materials identify external data-center partners. Neither source reports worker conditions or audit outcomes for data annotation, model evaluation, or partner-site operations.", "sourceUrls": [ "https://www.together.ai/careers", "https://www.together.ai/blog/together-ai-expands-in-europe" ], "confidence": "low", "unknowns": "Current pay, worker locations, exposure safeguards, and independent labor outcomes for Together or its infrastructure partners are not disclosed.", "suggestedScore": null, "scoreReason": "No score: public recruitment and infrastructure partnership information does not establish labor outcomes for data workers or contractors supporting inference.", "assessmentType": "argument", "argument": "Together publishes some employee benefit information and names infrastructure partners, but the cited materials give no annotation-worker, subcontractor, or partner-site outcomes. Request current vendor-level pay, location, exposure protections, audits, and remediation; company hiring benefits do not establish conditions across the inference supply chain.", "score": null, "href": "/providers/together-ai/#labour" }, "military-contracts": { "summary": "Together's terms describe hosted model/API services and its public materials describe commercial infrastructure expansion, but the reviewed sources do not identify a specific defense customer, award, or deployment for this provider. A cloud footprint or government eligibility alone would not establish military use.", "sourceUrls": [ "https://www.together.ai/terms-of-service", "https://www.together.ai/ai-factory" ], "confidence": "low", "unknowns": "These public documents are not a complete customer or contract register and cannot rule out undisclosed government work.", "suggestedScore": null, "scoreReason": "No score: the cited terms and infrastructure page establish service and capacity scope but disclose no defense contract or endpoint deployment.", "assessmentType": "argument", "argument": "Together's terms cover hosted APIs and its infrastructure page gives capacity and regional options, but neither identifies defense customers or model deployments. Leave the relationship unscored until a named contract or endpoint scope is documented; do not infer military use from infrastructure scale or general government eligibility.", "score": null, "href": "/providers/together-ai/#military-contracts" }, "israel-palestine": { "summary": "We did not find Together Computer, Inc. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Together Computer, Inc. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/providers/together-ai/#israel-palestine" } }, "sources": [ { "url": "https://www.together.ai/privacy", "title": "Together AI Privacy Policy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.together.ai/blog/together-ai-expands-in-europe", "title": "Bringing 100,000 GPUs to Europe", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.together.ai/models", "title": "Together AI models", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.together.ai/careers", "title": "Careers at Together AI", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.together.ai/ai-factory", "title": "Together AI Frontier AI Factory", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.together.ai/terms-of-service", "title": "Together AI Terms of Service", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/together-ai/", "overall": { "score": 6, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 10 } }, { "id": "fireworks-ai", "name": "Fireworks AI", "category": "providers", "organization": "Fireworks.ai, Inc.", "website": "https://fireworks.ai/", "summary": "Inference and fine-tuning platform hosting open-weight and other models; data and model terms depend on feature and model.", "scope": "Host/API operator, not necessarily original model developer; content handling and model corpus rights are distinct.", "dimensions": { "energy": { "summary": "Fireworks describes a high-volume managed inference platform, but its product disclosure does not allocate electricity, emissions, or water to a model, request, or serving region. The source supports the API service boundary, not a workload footprint.", "sourceUrls": [ "https://fireworks.ai/product/inference" ], "confidence": "low", "unknowns": "Facility, region, hardware, power mix, water, and per-request measurements are not disclosed in the product source.", "suggestedScore": null, "scoreReason": "No score: the service description provides no energy or emissions measurement that can be tied to an inference workload.", "assessmentType": "argument", "argument": "Fireworks publishes product information about managed, high-volume inference without a request-, model-, or region-level footprint. Ask for usage-linked power, carbon, and water data for the selected endpoint and location; platform scale or throughput claims cannot substitute for a service-boundary measurement.", "score": null, "href": "/providers/fireworks-ai/#energy" }, "training-rights": { "summary": "Provider’s commitment not to train on customer content does not establish licensing or consent for the hosted models’ original training data.", "sourceUrls": [ "https://trust.fireworks.ai/faq" ], "confidence": "low", "unknowns": "Model-specific provenance/rights vary and remain unresolved here.", "suggestedScore": null, "scoreReason": "Customer prompt protections do not establish rights in each hosted model’s upstream training corpus.", "assessmentType": "argument", "argument": "Provider’s commitment not to train on customer content does not establish licensing or consent for the hosted models’ original training data. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/fireworks-ai/#training-rights" }, "privacy": { "summary": "Fireworks states prompts, inputs, outputs and request content are not used to train its foundation models; its Trust Center describes ZDR for inference while stateful APIs and customer fine-tuning have separate handling.", "sourceUrls": [ "https://fireworks.ai/privacy-policy", "https://trust.fireworks.ai/faq" ], "confidence": "medium", "unknowns": "Check the data-handling page and stateful endpoint settings for a given product/configuration.", "suggestedScore": 8, "scoreReason": "Fireworks states inference ZDR and no model-training use, although stateful APIs, tuning, and account configurations have separate handling.", "assessmentType": "conditional", "argument": "Fireworks states inference ZDR and no model-training use, although stateful APIs, tuning, and account configurations have separate handling. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 8, "href": "/providers/fireworks-ai/#privacy" }, "openness": { "summary": "Fireworks hosts open-weight models and offers deployment options, but the inference platform itself is managed and each model’s license controls what can be moved.", "sourceUrls": [ "https://fireworks.ai/product/inference" ], "confidence": "medium", "unknowns": "The reviewed page does not establish that every advertised model is downloadable or that Fireworks platform code is open.", "suggestedScore": 5, "scoreReason": "Users can choose open models, but the hosted inference layer is not itself open or portable.", "assessmentType": "conditional", "argument": "Users can choose open models, but the hosted inference layer is not itself open or portable. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 5, "href": "/providers/fireworks-ai/#openness" }, "labour": { "summary": "Fireworks' public careers page lists roles across cloud infrastructure, AI training infrastructure, and model engineering, but it does not disclose pay or worker outcomes for labeling, moderation, or external contractors.", "sourceUrls": [ "https://fireworks.ai/careers" ], "confidence": "low", "unknowns": "Supplier standards, annotation vendor locations, exposure protections, pay, audits, and remediation outcomes were not established.", "suggestedScore": null, "scoreReason": "No score: job listings identify company roles, not labor conditions across the model-data and inference supply chains.", "assessmentType": "argument", "argument": "Fireworks' hiring page confirms cloud and AI-training infrastructure teams, but gives no worker-level conditions for data annotation or suppliers. Request independent vendor-level pay, location, exposure, audit, and remediation records before treating the absence of reported concerns as evidence of safeguards.", "score": null, "href": "/providers/fireworks-ai/#labour" }, "military-contracts": { "summary": "Fireworks says it can be purchased through Microsoft Foundry and cloud marketplaces and can run in customer environments such as AWS VPCs. These are distribution and hosting options, not evidence that a specific defense customer uses Fireworks inference.", "sourceUrls": [ "https://fireworks.ai/partners", "https://fireworks.ai/product/inference" ], "confidence": "low", "unknowns": "No named defense award or model/endpoint deployment is established by the cited partner and product pages; undisclosed contracts remain possible.", "suggestedScore": null, "scoreReason": "No score: channel partnerships and VPC options do not establish a defense contract or operational deployment.", "assessmentType": "argument", "argument": "Fireworks documents marketplace partnerships and customer-cloud deployment choices, but the sources do not identify a defense contract or Fireworks model used in a military workflow. Keep this unscored pending a named award and service scope; a hyperscaler relationship does not transfer that hyperscaler's defense use to Fireworks.", "score": null, "href": "/providers/fireworks-ai/#military-contracts" }, "israel-palestine": { "summary": "We did not find Fireworks.ai, Inc. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Fireworks.ai, Inc. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/providers/fireworks-ai/#israel-palestine" } }, "sources": [ { "url": "https://fireworks.ai/privacy-policy", "title": "Fireworks Privacy Policy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://trust.fireworks.ai/faq", "title": "Fireworks AI Trust Center FAQ", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://fireworks.ai/product/inference", "title": "Fireworks AI inference product", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://fireworks.ai/careers", "title": "Fireworks Careers", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://fireworks.ai/partners", "title": "Fireworks Partners", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/fireworks-ai/", "overall": { "score": 6.5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 4 } }, { "id": "groq", "name": "GroqCloud", "category": "providers", "organization": "Groq, Inc.", "website": "https://groq.com/", "summary": "Inference API operator offering hosted models; customer data controls do not determine model training corpus rights.", "scope": "Groq operates inference service and may host models developed by others; this record concerns GroqCloud API.", "dimensions": { "energy": { "summary": "Groq has published company claims of lower LPU power use and proposed a Norway inference center; its later DOE MOU says the parties would explore power-performance benchmarks. These are infrastructure and comparative-efficiency disclosures, not verified electricity, emissions, or water per GroqCloud request.", "sourceUrls": [ "https://groq.com/newsroom/groq-and-earth-wind-power-to-build-ai-compute-center-for-europe-in-norway-that-may-rival-tech-giant-scale", "https://groq.com/newsroom/groq-partners-with-us-department-of-energy-to-advance-ai-inference-and-next-generation-computing-infrastructure" ], "confidence": "low", "unknowns": "No independently verified API-level power, region-specific grid mix, water, or per-token/request footprint is provided.", "suggestedScore": null, "scoreReason": "No score: infrastructure efficiency claims are not tied to the live API route or a verified model/request boundary.", "assessmentType": "argument", "argument": "Groq reports an LPU efficiency comparison and a DOE collaboration to explore power benchmarks, while its Norway announcement described planned capacity. None supplies a verified footprint for a live GroqCloud model, route, or request. Ask for measured energy and water by serving region and workload.", "score": null, "href": "/providers/groq/#energy" }, "training-rights": { "summary": "Groq says customer inputs and outputs are not used to train models; this applies to customer data and does not disclose rights in upstream model training corpora.", "sourceUrls": [ "https://console.groq.com/docs/your-data" ], "confidence": "high", "unknowns": "Underlying model-specific training provenance and licenses remain open.", "suggestedScore": null, "scoreReason": "Customer input/output terms do not establish upstream corpus rights for the hosted models.", "assessmentType": "conditional", "argument": "Groq says customer inputs and outputs are not used to train models; this applies to customer data and does not disclose rights in upstream model training corpora. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/groq/#training-rights" }, "privacy": { "summary": "Groq documents no default retention of inference requests, up to 30-day temporary reliability/abuse logs, and customer-enabled ZDR; data for some GCP storage and persistence features is handled separately.", "sourceUrls": [ "https://console.groq.com/docs/your-data" ], "confidence": "high", "unknowns": "Confirm endpoint, persistence features, data-control settings, and region before use.", "suggestedScore": 7, "scoreReason": "Inference is not retained by default and ZDR is available, but temporary reliability/abuse logs may persist up to 30 days.", "assessmentType": "evidence", "argument": "Inference is not retained by default and ZDR is available, but temporary reliability/abuse logs may persist up to 30 days. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 7, "href": "/providers/groq/#privacy" }, "openness": { "summary": "Groq publishes its model list, including open-weight models, but its API is a hosted service and does not provide a portable serving stack.", "sourceUrls": [ "https://console.groq.com/docs/models" ], "confidence": "high", "unknowns": "Model availability changes; service access does not establish rights to weights, training data, or self-hosting for each model.", "suggestedScore": 4, "scoreReason": "Some weights are open upstream; Groq’s API and runtime remain hosted and closed.", "assessmentType": "conditional", "argument": "Some weights are open upstream; Groq’s API and runtime remain hosted and closed. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 4, "href": "/providers/groq/#openness" }, "labour": { "summary": "Groq's public careers link lists openings, and its service agreement defines GroqCloud and AI Model Services, but neither source reports conditions for data workers, contractors, or data-center supply-chain workers.", "sourceUrls": [ "https://jobs.gem.com/groq", "https://console.groq.com/docs/legal/services-agreement" ], "confidence": "low", "unknowns": "Worker pay, annotation locations, exposure controls, contractor audits, and remediation outcomes were not established.", "suggestedScore": null, "scoreReason": "No score: employment and service terms do not provide worker-level outcomes for the inference supply chain.", "assessmentType": "argument", "argument": "Groq's public hiring and service-contract pages identify its workforce and covered API services, but disclose no pay, exposure safeguards, or independent outcomes for labeling and infrastructure contractors. Request current vendor-level worker data and audits; a product agreement is not evidence of labor protections.", "score": null, "href": "/providers/groq/#labour" }, "military-contracts": { "summary": "Groq's Carahsoft agreement made GroqCloud inference available through federal procurement channels including Army ITES-SW2; a later DOE MOU concerns scientific discovery, energy-efficient inference, and Genesis Mission exploration. These establish public-sector access and a federal research relationship, not a named military model deployment.", "sourceUrls": [ "https://groq.com/newsroom/groq-and-carahsoft-partner-to-provide-rapid-ai-inference-speed-to-the-public-sector", "https://groq.com/newsroom/groq-partners-with-us-department-of-energy-to-advance-ai-inference-and-next-generation-computing-infrastructure", "https://console.groq.com/docs/legal/services-agreement" ], "confidence": "low", "unknowns": "No named defense award, operational military endpoint, or independent use/safeguard review is established; procurement-channel access does not prove a purchase.", "suggestedScore": null, "scoreReason": "No score: the sources document procurement access and DOE collaboration, but not an awarded defense task or actual military deployment of GroqCloud.", "assessmentType": "conditional", "argument": "Groq announced access through public-sector reseller and procurement vehicles, including an Army IT contract channel, and a separate DOE MOU for scientific inference work. Those are meaningful government relationships but do not establish an awarded military deployment or specific use; request the order, customer agency, and model endpoint before scoring operational use.", "score": null, "href": "/providers/groq/#military-contracts" }, "israel-palestine": { "summary": "We did not find Groq, Inc. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Groq, Inc. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/providers/groq/#israel-palestine" } }, "sources": [ { "url": "https://console.groq.com/docs/your-data", "title": "Your Data in GroqCloud", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://console.groq.com/docs/models", "title": "GroqCloud supported models", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://groq.com/newsroom/groq-and-earth-wind-power-to-build-ai-compute-center-for-europe-in-norway-that-may-rival-tech-giant-scale", "title": "Groq and Earth Wind & Power AI Compute Center announcement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://groq.com/newsroom/groq-partners-with-us-department-of-energy-to-advance-ai-inference-and-next-generation-computing-infrastructure", "title": "Groq and U.S. Department of Energy MOU", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://groq.com/newsroom/groq-and-carahsoft-partner-to-provide-rapid-ai-inference-speed-to-the-public-sector", "title": "Groq and Carahsoft public-sector distribution partnership", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://jobs.gem.com/groq", "title": "Groq Careers", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://console.groq.com/docs/legal/services-agreement", "title": "Groq Services Agreement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/groq/", "overall": { "score": 5.5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 12 } }, { "id": "cerebras", "name": "Cerebras Inference", "category": "providers", "organization": "Cerebras Systems, Inc.", "website": "https://www.cerebras.ai/inference", "summary": "Inference-as-a-service runs models on Cerebras infrastructure; model provider terms may govern output and model-specific rights.", "scope": "Cerebras is hardware/infrastructure and inference service operator; specific hosted model is third-party in some API uses.", "dimensions": { "energy": { "summary": "Cerebras announced a Finland data center scaling to 165 MW, with closed-loop cooling and planned heat recovery; this is infrastructure capacity and design information, not actual electricity/water use per Cerebras Inference request.", "sourceUrls": [ "https://investors.cerebras.ai/news-releases/news-release-details/cerebras-and-compute-nordic-finland-announce-new-165-mw-ai-data" ], "confidence": "low", "unknowns": "The company announcement is forward-looking; it does not disclose operational utilization, regional grid emissions, water data, or model/request allocation.", "suggestedScore": null, "scoreReason": "No score: a large planned facility and efficiency design do not quantify the footprint of Cerebras Cloud/API serving.", "assessmentType": "argument", "argument": "Cerebras disclosed a substantial Finland capacity plan and design measures for water reuse and heat recovery, but not metered operations or API-level allocation. Request the selected endpoint's region, utilization, energy, and water reporting; do not attribute all facility capacity or promised efficiency to a particular inference request.", "score": null, "href": "/providers/cerebras/#energy" }, "training-rights": { "summary": "Cerebras’ terms defer API output ownership to third-party model terms and disclaim Cerebras ownership; they do not establish source-data rights for underlying models.", "sourceUrls": [ "https://cloud.cerebras.ai/terms" ], "confidence": "medium", "unknowns": "Identify each model developer/license and investigate corpus provenance independently.", "suggestedScore": null, "scoreReason": "Terms defer to third-party models and do not establish corpus provenance or rights.", "assessmentType": "conditional", "argument": "Cerebras’ terms defer API output ownership to third-party model terms and disclaim Cerebras ownership; they do not establish source-data rights for underlying models. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/cerebras/#training-rights" }, "privacy": { "summary": "Cerebras’ privacy notice states inputs/outputs for its training, inference and chatbot services are not retained as described in that policy; API third-party model terms also apply.", "sourceUrls": [ "https://www.cerebras.ai/privacy-policy", "https://cloud.cerebras.ai/terms" ], "confidence": "medium", "unknowns": "Exact API/model, retention exceptions, processor/subprocessor terms and region need review.", "suggestedScore": 7, "scoreReason": "The company says inputs/outputs are not retained under its policy, but downstream third-party model terms still apply.", "assessmentType": "conditional", "argument": "The company says inputs/outputs are not retained under its policy, but downstream third-party model terms still apply. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 7, "href": "/providers/cerebras/#privacy" }, "openness": { "summary": "Cerebras offers hosted inference for selected models and an enterprise route for custom weights; its serving stack is not shown as portable.", "sourceUrls": [ "https://www.cerebras.ai/inference" ], "confidence": "medium", "unknowns": "The product page does not establish open-source Cerebras serving software, model training-data provenance, or availability of custom-weight deployment to all customers.", "suggestedScore": 4, "scoreReason": "The catalog includes open models, but the API runtime is managed and portability is limited.", "assessmentType": "conditional", "argument": "The catalog includes open models, but the API runtime is managed and portability is limited. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 4, "href": "/providers/cerebras/#openness" }, "labour": { "summary": "Cerebras' Business Partner Code sets expectations on wages, working hours, safety, forced labor, and human rights for suppliers and contractors. It requires records but the reviewed source does not report labor audits, worker outcomes, or model-data annotation conditions.", "sourceUrls": [ "https://www.cerebras.ai/business-partner-code-of-conduct" ], "confidence": "low", "unknowns": "No supplier audit findings, remediation outcomes, annotation vendor locations, or worker-pay data are disclosed in the cited code.", "suggestedScore": null, "scoreReason": "No score: the supplier code contains meaningful requirements but provides no evidence of implementation or inference-specific worker outcomes.", "assessmentType": "conditional", "argument": "Cerebras has a written business-partner code covering labor standards, but this is a commitment and recordkeeping requirement rather than a public audit result. Ask for recent supplier assessments, remediation closure data, and annotation-worker pay/location/exposure information before treating the code as proof of better working conditions.", "score": null, "href": "/providers/cerebras/#labour" }, "military-contracts": { "summary": "Cerebras says it joined the Department of Defense's Tradewinds Solutions Marketplace and partnered with Carahsoft to make its AI infrastructure available to federal agencies and defense organizations. The announcement establishes a procurement route, not an awarded deployment, named program, or specific inference workload.", "sourceUrls": [ "https://www.cerebras.ai/press-release/cerebras-partners-with-carahsoft-and-joins-department-of-defense-s-tradewinds-solutions" ], "confidence": "low", "unknowns": "The public announcement does not identify a military contract award, actual customer, operational endpoint, or safeguards review.", "suggestedScore": null, "scoreReason": "No score: DoD marketplace inclusion and reseller availability establish access to procurement, not a specific defense contract or use.", "assessmentType": "conditional", "argument": "Cerebras documents entry into a DoD procurement marketplace and a government reseller channel, but does not identify a specific award or operational use of Cerebras Inference. Request the award, customer, endpoint, and use restrictions; marketplace eligibility should not be presented as deployment evidence.", "score": null, "href": "/providers/cerebras/#military-contracts" }, "israel-palestine": { "summary": "We did not find Cerebras Systems, Inc. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Cerebras Systems, Inc. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/providers/cerebras/#israel-palestine" } }, "sources": [ { "url": "https://www.cerebras.ai/privacy-policy", "title": "Cerebras Privacy Policy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cloud.cerebras.ai/terms", "title": "Cerebras Cloud Terms of Use", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.cerebras.ai/inference", "title": "Cerebras inference", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://investors.cerebras.ai/news-releases/news-release-details/cerebras-and-compute-nordic-finland-announce-new-165-mw-ai-data", "title": "Cerebras and Compute Nordic Finland announce 165 MW AI data centre", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.cerebras.ai/business-partner-code-of-conduct", "title": "Cerebras Business Partner Code of Conduct", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.cerebras.ai/press-release/cerebras-partners-with-carahsoft-and-joins-department-of-defense-s-tradewinds-solutions", "title": "Cerebras joins DoD Tradewinds Solutions Marketplace", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/cerebras/", "overall": { "score": 5.5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 11 } }, { "id": "huggingface-inference-endpoints", "name": "Hugging Face Inference Endpoints", "category": "providers", "organization": "Hugging Face, Inc.", "website": "https://huggingface.co/inference-endpoints", "summary": "Managed dedicated infrastructure for deploying customer-selected Hub models; underlying model weights and licenses are user/model-repository specific.", "scope": "Hugging Face manages endpoint infrastructure; selected model may be third-party. Record excludes shared Inference Providers unless explicitly routed.", "dimensions": { "energy": { "summary": "Hugging Face describes Inference Endpoints as dedicated managed infrastructure on a cloud provider chosen by the customer, and separately recommends recording hardware, location, and emissions in model cards. The product docs do not provide an endpoint-level energy, emissions, or water measurement.", "sourceUrls": [ "https://huggingface.co/docs/inference-endpoints/en/about", "https://huggingface.co/docs/hub/main/en/model-cards-co2" ], "confidence": "low", "unknowns": "The chosen cloud, region, GPU utilization, and emissions may vary by deployment; no per-request or account-level energy/water report is established in these sources.", "suggestedScore": null, "scoreReason": "No score: customer-selected infrastructure and model-card guidance do not provide a measured footprint for the managed Endpoint.", "assessmentType": "argument", "argument": "Hugging Face lets customers select managed cloud infrastructure and encourages model-level carbon metadata, but that guidance is not an Inference Endpoint service footprint. Request metered energy, region-specific grid emissions, and water for the exact endpoint; do not treat a model card's training estimate as serving impact.", "score": null, "href": "/providers/huggingface-inference-endpoints/#energy" }, "training-rights": { "summary": "Endpoint product lets customers select model repositories and revision; Hugging Face states it does not own models/data uploaded. Corpus rights depend on selected model and cannot be inferred from endpoint hosting.", "sourceUrls": [ "https://huggingface.co/docs/inference-endpoints/en/about", "https://huggingface.co/docs/inference-endpoints/security" ], "confidence": "medium", "unknowns": "Review selected repository license, provenance, model card, and rights evidence.", "suggestedScore": null, "scoreReason": "Endpoint hosting controls do not establish upstream rights in customer-selected Hub models.", "assessmentType": "conditional", "argument": "Endpoint product lets customers select model repositories and revision; Hugging Face states it does not own models/data uploaded. Corpus rights depend on selected model and cannot be inferred from endpoint hosting. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/huggingface-inference-endpoints/#training-rights" }, "privacy": { "summary": "Hugging Face states it does not store endpoint payloads/tokens but stores logs for 30 days; offers public, protected, or private endpoints, with AWS/Azure/GCP infrastructure choices and regions.", "sourceUrls": [ "https://huggingface.co/docs/inference-endpoints/security" ], "confidence": "high", "unknowns": "Underlying cloud operator’s terms and selected endpoint configuration/region remain part of the data path.", "suggestedScore": 8, "scoreReason": "Endpoints provide private/protected deployment choices and state payloads are not stored, while logs persist for 30 days and cloud infrastructure is selectable.", "assessmentType": "conditional", "argument": "Endpoints provide private/protected deployment choices and state payloads are not stored, while logs persist for 30 days and cloud infrastructure is selectable. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 8, "href": "/providers/huggingface-inference-endpoints/#privacy" }, "openness": { "summary": "You choose the Hub model and revision and can deploy open models on selected cloud infrastructure; the managed endpoint runtime is still operated for you.", "sourceUrls": [ "https://huggingface.co/docs/inference-endpoints/en/about" ], "confidence": "high", "unknowns": "Model-specific license, weights availability, source, and documentation vary.", "suggestedScore": 7, "scoreReason": "Model and revision choice plus cloud selection gives substantial control, though the endpoint runtime remains managed.", "assessmentType": "evidence", "argument": "Model and revision choice plus cloud selection gives substantial control, though the endpoint runtime remains managed. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 7, "href": "/providers/huggingface-inference-endpoints/#openness" }, "labour": { "summary": "The 2024 Stanford Foundation Model Transparency Index reports that BigCode/Hugging Face/ServiceNow said it did not use external data labor for its flagship model. That is a model-development disclosure; Hugging Face's separate Endpoint product docs do not report labor conditions for endpoint operations or contractors.", "sourceUrls": [ "https://crfm.stanford.edu/fmti/May-2024/paper.pdf", "https://huggingface.co/docs/inference-endpoints/en/about" ], "confidence": "low", "unknowns": "The model-development finding is historical and does not establish current data labor practices, infrastructure-worker conditions, or supplier outcomes for the inference service.", "suggestedScore": null, "scoreReason": "No score: one upstream model disclosure does not establish labor conditions for Hugging Face's present managed inference service.", "assessmentType": "conditional", "argument": "Stanford's 2024 assessment records a no-external-data-labor disclosure for BigCode/Hugging Face/ServiceNow's flagship model, while Endpoint docs describe a separate managed hosting product. That is scope evidence, not a current worker-outcome audit of endpoint operations. Request current supplier and worker data for the selected hosting route.", "score": null, "href": "/providers/huggingface-inference-endpoints/#labour" }, "military-contracts": { "summary": "Hugging Face's Endpoint documentation describes customer-selected managed cloud deployment, but the reviewed sources do not identify a defense contract or military deployment of this product. Public models or government accounts on the Hub are not evidence that Hugging Face hosts a military inference endpoint.", "sourceUrls": [ "https://huggingface.co/docs/inference-endpoints/en/about", "https://huggingface.co/docs/inference-endpoints/security" ], "confidence": "low", "unknowns": "No named defense award or product-specific government deployment was established; these product pages are not a complete customer register.", "suggestedScore": null, "scoreReason": "No score: the source documents deployment architecture but no defense contract or military endpoint use.", "assessmentType": "argument", "argument": "Hugging Face documents managed endpoints on customer-selected cloud infrastructure, not customer-specific deployments. The reviewed evidence does not establish a military contract or endpoint use; request the named award and deployment scope before drawing a conclusion, and do not infer use from public Hub content.", "score": null, "href": "/providers/huggingface-inference-endpoints/#military-contracts" }, "israel-palestine": { "summary": "We did not find Hugging Face, Inc. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Hugging Face, Inc. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Hugging Face documents AWS and Azure infrastructure for private endpoints, linking those setups to named BDS targets. Check the actual cloud selection and billing arrangement; this does not classify Hugging Face itself as an official target.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://huggingface.co/docs/inference-endpoints/security" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": { "label": "Cloud choice matters", "sourceIds": [ "hf" ], "argument": "Hugging Face documents AWS and Azure infrastructure for private endpoints, linking those setups to named BDS targets. Check the actual cloud selection and billing arrangement; this does not classify Hugging Face itself as an official target." }, "businessContext": null, "score": null, "href": "/providers/huggingface-inference-endpoints/#israel-palestine" } }, "sources": [ { "url": "https://huggingface.co/docs/inference-endpoints/security", "title": "Hugging Face Inference Endpoints Security & Compliance", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/docs/inference-endpoints/en/about", "title": "About Inference Endpoints", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://huggingface.co/docs/hub/main/en/model-cards-co2", "title": "Hugging Face model-card carbon footprint guidance", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://crfm.stanford.edu/fmti/May-2024/paper.pdf", "title": "Foundation Model Transparency Index v1.1 (May 2024)", "kind": "research", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/huggingface-inference-endpoints/", "overall": { "score": 7.5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 2 } }, { "id": "replicate", "name": "Replicate", "category": "providers", "organization": "Replicate, LLC", "website": "https://replicate.com/", "summary": "Hosted API and cloud execution for Replicate and third-party models; each model’s license and provenance remain distinct.", "scope": "Replicate operates a model marketplace/API; models may be third-party offerings, and associated provider terms govern some rights.", "dimensions": { "energy": { "summary": "Replicate's terms describe managed model runs and configurable deployments, including customer choices of hardware and scaling, but do not identify serving-facility energy, emissions, or water. Hardware choice is useful context, not a usage-linked footprint.", "sourceUrls": [ "https://replicate.com/terms", "https://replicate.com/docs/topics/models/official-models" ], "confidence": "low", "unknowns": "Model-specific hardware and deployment locations are not tied to measured per-request electricity, emissions, or water in the cited terms/docs.", "suggestedScore": null, "scoreReason": "No score: hardware/deployment controls do not constitute model- or request-level energy reporting.", "assessmentType": "argument", "argument": "Replicate lets customers deploy models with hardware and scaling choices, but its public terms and model catalog do not provide measured power or water for an actual run. Request region-specific, workload-linked footprint data for the selected model and deployment; do not substitute GPU type or runtime for measurement.", "score": null, "href": "/providers/replicate/#energy" }, "training-rights": { "summary": "Terms define Marketplace Models as third-party offerings and point to separate Third Party Terms; service terms do not establish training corpus permissions for every model.", "sourceUrls": [ "https://replicate.com/terms" ], "confidence": "medium", "unknowns": "Model-by-model license and dataset provenance require review.", "suggestedScore": null, "scoreReason": "Marketplace models carry third-party terms; no catalog-wide corpus-rights evidence was established.", "assessmentType": "conditional", "argument": "Terms define Marketplace Models as third-party offerings and point to separate Third Party Terms; service terms do not establish training corpus permissions for every model. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/replicate/#training-rights" }, "privacy": { "summary": "Replicate’s API documentation says prediction inputs, outputs, files, and logs are automatically removed after one hour by default; predictions made through the web interface are kept indefinitely. Its general privacy policy permits retention as needed for service, legal, abuse-prevention, and claims purposes and does not establish a blanket no-training promise for every hosted model.", "sourceUrls": [ "https://replicate.com/privacy/", "https://replicate.com/terms", "https://replicate.com/privacy", "https://replicate.com/docs/topics/predictions/data-retention/" ], "confidence": "medium", "unknowns": "Confirm whether the user is calling the API or web interface, whether the model or feature has additional handling, and whether a customer contract changes backup, security, or legal-retention periods.", "suggestedScore": 6, "scoreReason": "The API has an explicit one-hour default cleanup for prediction content, but web predictions are retained indefinitely and the general privacy policy does not establish a universal model-training prohibition.", "assessmentType": "conditional", "argument": "A mid-range privacy judgment is supported for API predictions because content is removed after an hour by default, while web predictions persist and the general policy preserves service, legal, and abuse-prevention retention. Use the API only after checking the model and feature terms; do not assume the one-hour rule applies to web use or establishes a no-training guarantee.", "score": 6, "href": "/providers/replicate/#privacy" }, "openness": { "summary": "Replicate provides a stable API and model marketplace, but each model has separate terms and the hosted service does not make its runtime portable.", "sourceUrls": [ "https://replicate.com/docs/topics/models/official-models" ], "confidence": "low", "unknowns": "No claim can be made from silence; additional direct documentation or independently verified records are needed.", "suggestedScore": 4, "scoreReason": "The catalog enables model choice, but hosting and portability vary by model and the service remains managed.", "assessmentType": "conditional", "argument": "The catalog enables model choice, but hosting and portability vary by model and the service remains managed. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 4, "href": "/providers/replicate/#openness" }, "labour": { "summary": "Replicate's terms distinguish the platform from Marketplace Models and third-party model terms; its official-model directory shows that hosted models originate with external model developers. These sources do not report worker conditions for those developers, Replicate contractors, or data labeling.", "sourceUrls": [ "https://replicate.com/terms", "https://replicate.com/docs/topics/models/official-models" ], "confidence": "low", "unknowns": "No current provider-specific worker pay, location, exposure safeguards, audit, or remediation findings were established for the hosted catalog.", "suggestedScore": null, "scoreReason": "No score: third-party model authorship is explicit, but the sources provide no labor outcomes for the platform or upstream model workers.", "assessmentType": "argument", "argument": "Replicate's model directory and terms show that the hosted catalog includes third-party models with separate terms, so model-development labor cannot be attributed to Replicate or assumed to be covered by its API. The reviewed sources give no worker outcomes for either platform contractors or model teams; request disclosures from both the host and selected developer.", "score": null, "href": "/providers/replicate/#labour" }, "military-contracts": { "summary": "Replicate's current terms prohibit certain weapon-related and high-risk uses, including weapons-of-war development and weapons-system applications. This is a platform use restriction; the reviewed sources do not identify a defense contract or military deployment for Replicate's inference service.", "sourceUrls": [ "https://replicate.com/terms" ], "confidence": "low", "unknowns": "Terms do not establish how restrictions are enforced in a particular deployment or rule out undisclosed public-sector customers.", "suggestedScore": null, "scoreReason": "No score: the terms provide relevant use restrictions but do not establish a defense relationship or verified enforcement outcome.", "assessmentType": "conditional", "argument": "Replicate's terms restrict weapons-of-war and specified high-risk uses, which is a relevant contractual safeguard, but they do not document actual defense contracts or prove enforcement in a particular deployment. Treat actual deployments as unverified and request any named government award, service scope, and compliance evidence.", "score": null, "href": "/providers/replicate/#military-contracts" }, "israel-palestine": { "summary": "We did not find Replicate, LLC among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Replicate, LLC is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/providers/replicate/#israel-palestine" } }, "sources": [ { "url": "https://replicate.com/privacy/", "title": "Replicate Privacy Policy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://replicate.com/terms", "title": "Replicate Terms of Service", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://replicate.com/docs/topics/models/official-models", "title": "Official models - Replicate", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://replicate.com/privacy", "title": "Replicate privacy policy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://replicate.com/docs/topics/predictions/data-retention/", "title": "Data retention for Replicate predictions", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/replicate/", "overall": { "score": 5, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 15 } }, { "id": "deepinfra", "name": "DeepInfra", "category": "providers", "organization": "DeepInfra, Inc.", "website": "https://deepinfra.com/", "summary": "Inference API serving models from a catalog; input handling and underlying model rights are separate layers.", "scope": "DeepInfra operates hosted inference; selected models may be third-party/open-weight.", "dimensions": { "energy": { "summary": "DeepInfra describes operating GPU infrastructure across multiple U.S. data centers, but its public service materials do not allocate facility electricity, water, or emissions to a model, region, or request.", "sourceUrls": [ "https://deepinfra.com/about", "https://deepinfra.com/series-b" ], "confidence": "low", "unknowns": "Request-level electricity, emissions, and water use remain unknown.", "suggestedScore": null, "scoreReason": "Not rated: No service-level power, emissions, or water figures.", "assessmentType": "argument", "argument": "DeepInfra’s infrastructure disclosures establish that it operates GPU capacity, not the energy or water consumed by a particular inference route. Ask for a workload- and region-linked footprint before comparing deployments; corporate scale or efficiency claims cannot substitute for that boundary.", "score": null, "href": "/providers/deepinfra/#energy" }, "training-rights": { "summary": "DeepInfra’s terms leave third-party model rights to each provider’s license and disclaim responsibility for third-party content; no catalog-wide training-corpus provenance or rights audit is disclosed. The host’s no-training rule applies to customer data, not upstream model corpora.", "sourceUrls": [ "https://deepinfra.com/terms" ], "confidence": "high", "unknowns": "Underlying model corpus provenance, licenses and consent remain model-specific.", "suggestedScore": null, "scoreReason": "No-training customer-data terms do not establish upstream model-corpus rights.", "assessmentType": "conditional", "argument": "DeepInfra’s terms make model-specific licensing the customer’s responsibility and do not provide an audited corpus-rights ledger, so the host cannot establish permissions for every model it serves. Before use, inspect the selected model’s license and provenance separately; the host’s no-training promise protects your inputs, not the model’s training set.", "score": null, "href": "/providers/deepinfra/#training-rights" }, "privacy": { "summary": "DeepInfra’s terms, last modified October 15, 2026, promise zero data retention for customer data and prohibit using it to train or improve models, except as necessary to provide the service. They exclude customer-authorized support retention (deleted within 30 days after resolution), non-content metadata, legal/security/abuse records, and opted-in features such as batch, asynchronous outputs, third-party integrations, or paid cache; third-party-hosted models follow their own terms.", "sourceUrls": [ "https://deepinfra.com/privacy", "https://deepinfra.com/terms" ], "confidence": "high", "unknowns": "Confirm the chosen model is hosted by DeepInfra rather than a third party and avoid opting into batch, paid cache, or other persistence without reviewing that feature’s retention terms.", "suggestedScore": 8, "scoreReason": "Current terms expressly promise service-level ZDR and no customer-data training, with bounded, enumerated exceptions; the contract distinguishes third-party models and customer-selected storage features.", "assessmentType": "evidence", "argument": "The 8/10 rating rests on DeepInfra’s current terms: content is not retained after request processing or used to train models, with stated legal, security, support, and customer-selected-feature exceptions. Choose a DeepInfra-hosted model and review the selected feature’s terms; the contract explicitly sends third-party model data to the provider’s own policy.", "score": 8, "href": "/providers/deepinfra/#privacy" }, "openness": { "summary": "The catalog exposes a broad set of upstream open-weight models, but the service terms treat DeepInfra’s inference service as proprietary and preserve third-party model licensing separately. Access to open weights therefore does not establish that DeepInfra’s serving stack is inspectable or portable.", "sourceUrls": [ "https://deepinfra.com/terms", "https://deepinfra.com/models" ], "confidence": "medium", "unknowns": "The reviewed sources do not document a reproducible server/runtime implementation or a path to move the managed inference stack; model availability and license differ by entry.", "suggestedScore": null, "scoreReason": "Open-weight models can be selected, but the service’s terms and runtime remain provider-controlled, so the model catalog alone does not justify a service-openness number.", "assessmentType": "conditional", "argument": "The catalog confirms open-weight choices, while the service terms keep DeepInfra’s runtime under provider control and assign third-party model licensing separately. Leave the hosted-service openness score unrated until the serving stack’s inspectability and portability are documented; choose and self-host a specific model if those controls matter.", "score": null, "href": "/providers/deepinfra/#openness" }, "labour": { "summary": "DeepInfra’s public service and company materials describe inference infrastructure and customer-use terms, but do not report worker outcomes for data work, contractors, or data-center operations.", "sourceUrls": [ "https://deepinfra.com/about", "https://deepinfra.com/terms" ], "confidence": "low", "unknowns": "Conditions for contractors, annotators, and model evaluators remain unknown.", "suggestedScore": null, "scoreReason": "Unrated: the reviewed sources do not establish worker-level outcomes.", "assessmentType": "argument", "argument": "The available company materials bound the business as model hosting on GPU infrastructure but do not identify its labor suppliers, pay, working conditions, or audit results. Treat worker protections as unverified and ask for current contractor and facility-vendor disclosures before relying on a labor claim.", "score": null, "href": "/providers/deepinfra/#labour" }, "military-contracts": { "summary": "DeepInfra’s terms restrict certain weapons-system uses, but the reviewed company materials do not name a defense contract or establish an actual military deployment.", "sourceUrls": [ "https://deepinfra.com/terms", "https://deepinfra.com/about" ], "confidence": "low", "unknowns": "No inference about undisclosed customers or contracts is possible from silence.", "suggestedScore": null, "scoreReason": "Unrated: no relevant public defense deployment surfaced in the reviewed sources.", "assessmentType": "argument", "argument": "The use policy is a customer-facing restriction, not evidence about procurement or enforcement in a specific deployment. Keep the contract relationship unrated unless the provider identifies a customer, award, model or endpoint scope, and applicable controls.", "score": null, "href": "/providers/deepinfra/#military-contracts" }, "israel-palestine": { "summary": "We did not find DeepInfra, Inc. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "DeepInfra, Inc. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/providers/deepinfra/#israel-palestine" } }, "sources": [ { "url": "https://deepinfra.com/privacy", "title": "DeepInfra Privacy Policy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://deepinfra.com/terms", "title": "DeepInfra Terms of Service", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://deepinfra.com/models", "title": "DeepInfra model catalog", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://deepinfra.com/series-b", "title": "DeepInfra raises $107M Series B to scale inference infrastructure", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://deepinfra.com/about", "title": "About DeepInfra", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/deepinfra/", "overall": { "score": 8, "rated": 1, "total": 6, "coverage": 0.16666666666666666, "partial": true, "color": "green", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 1 } }, { "id": "openrouter", "name": "OpenRouter", "category": "providers", "organization": "OpenRouter, Inc.", "website": "https://openrouter.ai/", "summary": "Model access intermediary routes requests to selected or automatically routed model providers; privacy and training handling depends on provider/model settings.", "scope": "OpenRouter is an intermediary, not the inference host/model owner in every transaction; actual endpoint and model provider are essential context.", "dimensions": { "energy": { "summary": "OpenRouter describes itself as an interface that routes requests to model providers; its public materials do not report a request-level electricity, emissions, or water footprint for the router or downstream route.", "sourceUrls": [ "https://openrouter.ai/about/", "https://openrouter.ai/terms" ], "confidence": "low", "unknowns": "Request-level electricity, emissions, and water use remain unknown.", "suggestedScore": null, "scoreReason": "Not rated: No service-level power, emissions, or water figures.", "assessmentType": "argument", "argument": "A routed request’s facility and model can vary by selected provider, while OpenRouter’s public disclosures do not attach a measured footprint to that path. Compare only after obtaining the selected upstream provider, region, and usage-linked energy data.", "score": null, "href": "/providers/openrouter/#energy" }, "training-rights": { "summary": "OpenRouter’s terms/privacy explain intermediary routing and defer model corpus/data practices to model providers; no single OpenRouter policy establishes training rights for all catalog models.", "sourceUrls": [ "https://openrouter.ai/privacy/", "https://openrouter.ai/terms" ], "confidence": "high", "unknowns": "Model-provider-specific policies, model license and training corpus provenance are required.", "suggestedScore": null, "scoreReason": "As a router, OpenRouter defers corpus and upstream data practices to selected providers; no catalog-wide rights due diligence was established.", "assessmentType": "conditional", "argument": "OpenRouter’s terms/privacy explain intermediary routing and defer model corpus/data practices to model providers; no single OpenRouter policy establishes training rights for all catalog models. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/openrouter/#training-rights" }, "privacy": { "summary": "OpenRouter states it does not use inputs/outputs to train its own models, but transmits inputs to selected or automatically routed model providers, whose retention and training terms differ; uploaded files may be retained until deleted.", "sourceUrls": [ "https://openrouter.ai/privacy/" ], "confidence": "high", "unknowns": "Review provider-specific data practices, routing selection, account controls, and whether files/stateful features are used.", "suggestedScore": 6, "scoreReason": "OpenRouter says it does not train on prompts, but routes content to selected providers with differing practices and may retain uploaded files.", "assessmentType": "conditional", "argument": "OpenRouter says it does not train on prompts, but routes content to selected providers with differing practices and may retain uploaded files. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 6, "href": "/providers/openrouter/#privacy" }, "openness": { "summary": "OpenRouter lets users select models and providers and configure routing, but it forwards requests to other hosted services; its router is not a self-hostable model stack.", "sourceUrls": [ "https://openrouter.ai/" ], "confidence": "low", "unknowns": "No claim can be made from silence; additional direct documentation or independently verified records are needed.", "suggestedScore": 6, "scoreReason": "Model/provider choice and routing controls help portability, but every request still relies on another hosted provider.", "assessmentType": "conditional", "argument": "Model/provider choice and routing controls help portability, but every request still relies on another hosted provider. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 6, "href": "/providers/openrouter/#openness" }, "labour": { "summary": "OpenRouter’s careers page describes its internal team and hiring, while its routing terms place model delivery with separate providers; neither establishes labor conditions across those upstream model and infrastructure suppliers.", "sourceUrls": [ "https://openrouter.ai/careers", "https://openrouter.ai/terms" ], "confidence": "low", "unknowns": "Conditions for contractors, annotators, and model evaluators remain unknown.", "suggestedScore": null, "scoreReason": "Unrated: the reviewed sources do not establish worker-level outcomes.", "assessmentType": "argument", "argument": "The company’s own recruiting information is not evidence about annotation, data-center, or upstream-provider work. Request supplier-level labor and audit information for the actual model route; otherwise the worker conditions remain unverified.", "score": null, "href": "/providers/openrouter/#labour" }, "military-contracts": { "summary": "OpenRouter’s terms and routing description establish an intermediary role, but the reviewed sources do not identify a defense award or military deployment tied to an OpenRouter endpoint.", "sourceUrls": [ "https://openrouter.ai/terms", "https://openrouter.ai/about/" ], "confidence": "low", "unknowns": "No inference about undisclosed customers or contracts is possible from silence.", "suggestedScore": null, "scoreReason": "Unrated: no relevant public defense deployment surfaced in the reviewed sources.", "assessmentType": "argument", "argument": "A routing layer can expose models operated by separate providers, so neither its own terms nor silence about contracts establishes the downstream use of each route. Assess a named customer and model-provider path separately, with the endpoint and contractual restrictions specified.", "score": null, "href": "/providers/openrouter/#military-contracts" }, "israel-palestine": { "summary": "We did not find OpenRouter, Inc. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "OpenRouter, Inc. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. OpenRouter’s directory includes Google Vertex, Amazon Bedrock and Azure. Routes through those services involve named BDS targets; the directory does not mean every OpenRouter request uses them, so inspect and restrict your provider route.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "https://openrouter.ai/providers", "https://openrouter.ai/provider/azure", "https://openrouter.ai/provider/amazon-bedrock", "https://openrouter.ai/provider/google-vertex", "https://openrouter.ai/docs/guides/routing/provider-selection" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": { "label": "Route dependent", "sourceIds": [ "router", "router-azure", "router-aws", "router-google", "router-routing" ], "argument": "OpenRouter’s directory includes Google Vertex, Amazon Bedrock and Azure. Routes through those services involve named BDS targets; the directory does not mean every OpenRouter request uses them, so inspect and restrict your provider route." }, "businessContext": null, "score": null, "href": "/providers/openrouter/#israel-palestine" } }, "sources": [ { "url": "https://openrouter.ai/privacy/", "title": "OpenRouter Privacy Policy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openrouter.ai/terms", "title": "OpenRouter Terms of Service", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openrouter.ai/", "title": "OpenRouter - The Unified Interface For Every Model", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openrouter.ai/about/", "title": "About OpenRouter", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openrouter.ai/careers", "title": "Careers at OpenRouter", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://openrouter.ai/providers", "title": "OpenRouter: provider directory and available routes", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openrouter.ai/provider/azure", "title": "OpenRouter: Azure provider route", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openrouter.ai/provider/amazon-bedrock", "title": "OpenRouter: Amazon Bedrock provider route", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openrouter.ai/provider/google-vertex", "title": "OpenRouter: Google Vertex provider route", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://openrouter.ai/docs/guides/routing/provider-selection", "title": "OpenRouter: selecting and limiting provider routes", "kind": "primary", "checkedAt": "2026-10-23" } ], "href": "/providers/openrouter/", "overall": { "score": 6, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 8 } }, { "id": "scaleway", "name": "Scaleway Generative APIs", "category": "providers", "organization": "Scaleway SAS", "website": "https://www.scaleway.com/en/generative-apis/", "summary": "European managed serverless and dedicated inference offering. Serverless currently hosted in Paris; dedicated deployments allow customer selection of configuration.", "scope": "Scaleway hosts the selected models on its infrastructure; third-party model creators’ original data rights are distinct.", "dimensions": { "energy": { "summary": "Scaleway documentation identifies the Paris data center, selectable GPU instance types, and available GPU power/usage metrics for dedicated deployments; it does not provide power/emissions per serverless request.", "sourceUrls": [ "https://www.scaleway.com/en/docs/generative-apis/faq/" ], "confidence": "medium", "unknowns": "Corporate energy claims and facility-level values do not resolve individual model or request impact; no per-request estimate found.", "suggestedScore": 6, "scoreReason": "Dedicated deployments expose GPU type and power/usage metrics, but serverless request footprint and emissions remain undisclosed.", "assessmentType": "conditional", "argument": "Dedicated deployments expose GPU type and power/usage metrics, but serverless request footprint and emissions remain undisclosed. For a deployment, request or export the selected service and region’s usage-linked footprint when available; do not treat company-wide renewable matching as per-request impact.", "score": 6, "href": "/providers/scaleway/#energy" }, "training-rights": { "summary": "Scaleway says customer inputs are not used to train underlying base models. This hosting/API assurance does not establish legal rights in the hosted models’ training corpora.", "sourceUrls": [ "https://www.scaleway.com/en/docs/generative-apis/reference-content/data-privacy/" ], "confidence": "high", "unknowns": "Model-specific licensing, provenance and dataset consent/compensation remain unverified.", "suggestedScore": null, "scoreReason": "Scaleway’s customer-data policy does not establish the training corpus rights of selected hosted models.", "assessmentType": "conditional", "argument": "Scaleway says customer inputs are not used to train underlying base models. This hosting/API assurance does not establish legal rights in the hosted models’ training corpora. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/scaleway/#training-rights" }, "privacy": { "summary": "Scaleway says its Generative APIs default to ZDR, collect metadata, and may temporarily store full request content for rare service-harm/error investigations up to two weeks. Serverless currently runs in Paris; dedicated deployment is advised for single-region processing.", "sourceUrls": [ "https://www.scaleway.com/en/docs/generative-apis/reference-content/data-privacy/", "https://www.scaleway.com/en/docs/generative-apis/faq/" ], "confidence": "high", "unknowns": "Verify serverless/dedicated product, location, specific model terms, and temporary investigation exception.", "suggestedScore": 8, "scoreReason": "Default ZDR and Paris serverless hosting are disclosed; rare investigations can retain full request content for up to two weeks, and metadata is collected.", "assessmentType": "conditional", "argument": "Default ZDR and Paris serverless hosting are disclosed; rare investigations can retain full request content for up to two weeks, and metadata is collected. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 8, "href": "/providers/scaleway/#privacy" }, "openness": { "summary": "Scaleway offers serverless and dedicated deployments, including custom-model options, but its hosted API stack is not shown as portable.", "sourceUrls": [ "https://www.scaleway.com/en/docs/generative-apis/faq/" ], "confidence": "low", "unknowns": "No claim can be made from silence; additional direct documentation or independently verified records are needed.", "suggestedScore": 5, "scoreReason": "Dedicated deployments and model choice help; the serving stack is still managed.", "assessmentType": "conditional", "argument": "Dedicated deployments and model choice help; the serving stack is still managed. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 5, "href": "/providers/scaleway/#openness" }, "labour": { "summary": "Iliad’s 2025 Universal Registration Document describes group-level supplier evaluation, a Partner Relations Charter, and monitoring/corrective processes; it does not report worker-level outcomes for Generative APIs or its annotation and facility supply chain.", "sourceUrls": [ "https://www.iliad.fr/media/ILIAD_URD_30042026_ENG_834039a832.pdf" ], "confidence": "low", "unknowns": "Conditions for contractors, annotators, and model evaluators remain unknown.", "suggestedScore": null, "scoreReason": "Unrated: the reviewed sources do not establish worker-level outcomes.", "assessmentType": "argument", "argument": "The parent-group supplier controls are relevant policy evidence, but they do not show audit coverage or results for this API’s vendors. Ask Scaleway for the current supplier scope and worker-level findings before treating the group commitment as an endpoint safeguard.", "score": null, "href": "/providers/scaleway/#labour" }, "military-contracts": { "summary": "Scaleway announced an Airbus sovereign-cloud contract for selected enterprise applications, but the announcement does not connect that workload to Generative APIs or identify a military endpoint deployment.", "sourceUrls": [ "https://www.scaleway.com/en/news/scaleway-secures-european-trusted-cloud-services-contract-with-airbus/" ], "confidence": "low", "unknowns": "No inference about undisclosed customers or contracts is possible from silence.", "suggestedScore": null, "scoreReason": "Unrated: no relevant public defense deployment surfaced in the reviewed sources.", "assessmentType": "argument", "argument": "Airbus is an aerospace contractor, yet the disclosed contract is not evidence that this API supports military operations. Keep this dimension unrated absent a named defense award or endpoint scope; verify the workload rather than infer it from the customer’s sector.", "score": null, "href": "/providers/scaleway/#military-contracts" }, "israel-palestine": { "summary": "We did not find Scaleway SAS among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Scaleway SAS is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/providers/scaleway/#israel-palestine" } }, "sources": [ { "url": "https://www.scaleway.com/en/docs/generative-apis/reference-content/data-privacy/", "title": "Scaleway Generative APIs Privacy Policy", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.scaleway.com/en/docs/generative-apis/faq/", "title": "Scaleway Generative APIs FAQ", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.iliad.fr/media/ILIAD_URD_30042026_ENG_834039a832.pdf", "title": "iliad Group 2025 Universal Registration Document", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.scaleway.com/en/news/scaleway-secures-european-trusted-cloud-services-contract-with-airbus/", "title": "Scaleway selected as Airbus sovereign cloud provider", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/scaleway/", "overall": { "score": 6.3, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 5 } }, { "id": "infomaniak", "name": "Infomaniak AI Services", "category": "providers", "organization": "Infomaniak Network SA", "website": "https://www.infomaniak.com/en/hosting/ai-services", "summary": "Swiss-hosted managed API access to open-source models. Corporate sustainability and data-handling assertions are provider disclosures, not independent measurements.", "scope": "Infomaniak hosts models in its infrastructure; catalog model licenses/provenance remain distinct.", "dimensions": { "energy": { "summary": "Infomaniak’s impact report describes site-level efficiency and heat recovery alongside its renewable-powered AI service claim. These are company/data-center disclosures, not an independently verified allocation for an AI request or model; compare them as infrastructure context and ask for workload-level reporting.", "sourceUrls": [ "https://www.infomaniak.com/en/hosting/ai-services", "https://www.infomaniak.com/en/impact-report" ], "confidence": "medium", "unknowns": "No independent per-request energy/emissions/water figure located; scope and hourly electricity mix require verification.", "suggestedScore": 6, "scoreReason": "Infomaniak reports renewable-powered AI infrastructure and heat recovery, but provides no independently verified model-, request-, or region-specific footprint.", "assessmentType": "conditional", "argument": "Infomaniak reports renewable-powered AI infrastructure and heat recovery, but provides no independently verified model-, request-, or region-specific footprint. For a deployment, request or export the selected service and region’s usage-linked footprint when available; do not treat company-wide renewable matching as per-request impact.", "score": 6, "href": "/providers/infomaniak/#energy" }, "training-rights": { "summary": "Infomaniak says queries are not used to train models, but its terms disclaim liability for open-source models potentially trained on copyrighted data without authorization. Hosting/privacy claims do not clear corpus rights.", "sourceUrls": [ "https://www.infomaniak.com/en/hosting/ai-services", "https://welcome.infomaniak.com/api/components/cgu/latest?id=87&locale=en_GB" ], "confidence": "medium", "unknowns": "Specific model licenses, provenance and rights evidence need model-by-model review.", "suggestedScore": null, "scoreReason": "No-query-training commitments concern customer data and do not establish upstream corpus permissions; terms explicitly leave open-source model rights model-dependent.", "assessmentType": "conditional", "argument": "Infomaniak says queries are not used to train models, but its terms disclaim liability for open-source models potentially trained on copyrighted data without authorization. Hosting/privacy claims do not clear corpus rights. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/infomaniak/#training-rights" }, "privacy": { "summary": "Infomaniak says AI queries are not recorded or used to train models or improve services and describes processing in Switzerland/Europe. Specific contract terms should be checked for the product/API chosen.", "sourceUrls": [ "https://www.infomaniak.com/en/hosting/ai-services", "https://welcome.infomaniak.com/api/components/cgu/latest?id=87&locale=en_GB" ], "confidence": "medium", "unknowns": "Confirm actual service version, logging/metadata, account and enterprise contract terms.", "suggestedScore": 7, "scoreReason": "The company states queries are not logged or used for training and locates processing in Switzerland/Europe, but contract/product-specific details should be checked.", "assessmentType": "conditional", "argument": "The company states queries are not logged or used for training and locates processing in Switzerland/Europe, but contract/product-specific details should be checked. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 7, "href": "/providers/infomaniak/#privacy" }, "openness": { "summary": "Infomaniak serves open-source models through its managed infrastructure; users can choose models, but the serving stack is not shown as portable.", "sourceUrls": [ "https://www.infomaniak.com/en/hosting/ai-services" ], "confidence": "low", "unknowns": "No claim can be made from silence; additional direct documentation or independently verified records are needed.", "suggestedScore": 5, "scoreReason": "Users can choose open-source models, but only through Infomaniak’s managed service.", "assessmentType": "conditional", "argument": "Users can choose open-source models, but only through Infomaniak’s managed service. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 5, "href": "/providers/infomaniak/#openness" }, "labour": { "summary": "Infomaniak’s impact report describes employee policies and workplace practices, including working-time and wellbeing measures, but does not independently audit worker outcomes across AI model, annotation, and infrastructure suppliers.", "sourceUrls": [ "https://www.infomaniak.com/en/impact-report" ], "confidence": "low", "unknowns": "Conditions for contractors, annotators, and model evaluators remain unknown.", "suggestedScore": null, "scoreReason": "Unrated: the reviewed sources do not establish worker-level outcomes.", "assessmentType": "argument", "argument": "The report provides company-level employee evidence, so labor is not wholly opaque; however, its scope does not establish conditions across the AI service’s external supply chain or independent outcomes. Ask for current vendor coverage and audit results before extending internal policies to the full service.", "score": null, "href": "/providers/infomaniak/#labour" }, "military-contracts": { "summary": "Infomaniak markets sovereign-cloud services for public authorities and sensitive organizations, but the reviewed materials do not identify a defense contract or military deployment of its AI services.", "sourceUrls": [ "https://www.infomaniak.com/en/sovereign-cloud", "https://www.infomaniak.com/en/hosting/ai-services" ], "confidence": "low", "unknowns": "No inference about undisclosed customers or contracts is possible from silence.", "suggestedScore": null, "scoreReason": "Unrated: no relevant public defense deployment surfaced in the reviewed sources.", "assessmentType": "argument", "argument": "Public-sector suitability is not proof of military procurement or use. Keep the relationship unrated until a named contract identifies the AI service, customer, and operational scope; do not infer use from the customer segment alone.", "score": null, "href": "/providers/infomaniak/#military-contracts" }, "israel-palestine": { "summary": "We did not find Infomaniak Network SA among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Infomaniak Network SA is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/providers/infomaniak/#israel-palestine" } }, "sources": [ { "url": "https://www.infomaniak.com/en/hosting/ai-services", "title": "Infomaniak AI Services", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://welcome.infomaniak.com/api/components/cgu/latest?id=87&locale=en_GB", "title": "Infomaniak LLM API Terms and Conditions", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.infomaniak.com/en/impact-report", "title": "Impact report: For a responsible digital future and a sustainable cloud", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.infomaniak.com/en/sovereign-cloud", "title": "Infomaniak sovereign cloud", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/infomaniak/", "overall": { "score": 6, "rated": 3, "total": 6, "coverage": 0.5, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 6 } }, { "id": "ovhcloud-ai-endpoints", "name": "OVHcloud AI Endpoints", "category": "providers", "organization": "OVH Groupe SA", "website": "https://www.ovhcloud.com/en-gb/public-cloud/ai-endpoints/", "summary": "Serverless inference API for a catalog of open-weight models hosted by OVHcloud. Model license and upstream corpus provenance are distinct from hosting.", "scope": "OVHcloud operates API and infrastructure; selected catalog models may originate with third parties.", "dimensions": { "energy": { "summary": "OVHcloud publishes audited data-center environmental indicators and an Environmental Impact Tracker for listed cloud services, but its announcement does not establish an equivalent usage-linked footprint for AI Endpoints.", "sourceUrls": [ "https://corporate.ovhcloud.com/asia/sustainability/environment/", "https://corporate.ovhcloud.com/en-sg/newsroom/news/environmental-impact-tracker/" ], "confidence": "low", "unknowns": "Request-level electricity, emissions, and water use remain unknown.", "suggestedScore": null, "scoreReason": "Not rated: No service-level power, emissions, or water figures.", "assessmentType": "argument", "argument": "The corporate tracker can help assess covered infrastructure, yet the published product scope does not confirm model- or request-level AI Endpoints allocation. Check whether the selected endpoint is included before using those totals; otherwise request a service- and region-specific estimate.", "score": null, "href": "/providers/ovhcloud-ai-endpoints/#energy" }, "training-rights": { "summary": "Hosting service claims do not use customer prompts to train models; this does not substantiate licenses or provenance for model pretraining data.", "sourceUrls": [ "https://www.ovhcloud.com/en-gb/public-cloud/ai-endpoints/" ], "confidence": "low", "unknowns": "Review each model license and source-data disclosures independently.", "suggestedScore": null, "scoreReason": "Hosting-level no-training terms do not establish rights or provenance for open-weight model corpora.", "assessmentType": "argument", "argument": "Hosting service claims do not use customer prompts to train models; this does not substantiate licenses or provenance for model pretraining data. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/ovhcloud-ai-endpoints/#training-rights" }, "privacy": { "summary": "OVHcloud states it does not use customer data to train/improve models and describes zero data retention except billing data.", "sourceUrls": [ "https://www.ovhcloud.com/en-gb/public-cloud/ai-endpoints/" ], "confidence": "medium", "unknowns": "Check current product-specific terms, logs/abuse handling, region and any stateful/batch feature exceptions.", "suggestedScore": 7, "scoreReason": "OVHcloud states ZDR and no model-training use apart from billing records, but reviewed evidence is a company product claim without independent audit.", "assessmentType": "conditional", "argument": "OVHcloud states ZDR and no model-training use apart from billing records, but reviewed evidence is a company product claim without independent audit. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 7, "href": "/providers/ovhcloud-ai-endpoints/#privacy" }, "openness": { "summary": "OVHcloud offers an open-weight model catalog through its managed API; model weights may be open, but the service runtime is not.", "sourceUrls": [ "https://www.ovhcloud.com/en-gb/public-cloud/ai-endpoints/" ], "confidence": "low", "unknowns": "No claim can be made from silence; additional direct documentation or independently verified records are needed.", "suggestedScore": 5, "scoreReason": "The catalog offers open weights, but the managed API itself is not portable.", "assessmentType": "conditional", "argument": "The catalog offers open weights, but the managed API itself is not portable. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 5, "href": "/providers/ovhcloud-ai-endpoints/#openness" }, "labour": { "summary": "OVHcloud’s sustainability statement and group reporting describe company-wide workforce and supply-chain policies, but the reviewed material does not disclose worker outcomes specifically for AI Endpoints or model-data contractors.", "sourceUrls": [ "https://corporate.ovhcloud.com/sites/default/files/2025-11/sustainability_statement.pdf", "https://corporate.ovhcloud.com/en-gb/sustainability/" ], "confidence": "low", "unknowns": "Conditions for contractors, annotators, and model evaluators remain unknown.", "suggestedScore": null, "scoreReason": "Unrated: the reviewed sources do not establish worker-level outcomes.", "assessmentType": "argument", "argument": "Company-wide reporting is useful context but cannot establish conditions for the vendors that label, maintain, or operate this particular service. Ask for AI-service supplier coverage, worker grievance and remediation results, and independent audit scope before treating the policy as stronger assurance.", "score": null, "href": "/providers/ovhcloud-ai-endpoints/#labour" }, "military-contracts": { "summary": "OVHcloud’s FY2026 half-year results say it is developing a Defence vertical after approaches from European defence ministries and cite defense use cases; they do not name an AI Endpoints contract or deployment.", "sourceUrls": [ "https://corporate.ovhcloud.com/sites/default/files/external_files/2026-04-09-ovhcloud-h1-fy26-pr-eng_vdef_0.pdf" ], "confidence": "low", "unknowns": "No inference about undisclosed customers or contracts is possible from silence.", "suggestedScore": null, "scoreReason": "Unrated: no relevant public defense deployment surfaced in the reviewed sources.", "assessmentType": "argument", "argument": "This is concrete defense-sector commercial intent, not proof that the hosted AI endpoint is deployed for those use cases. The link to this product remains unverified until an award, customer, and AI service scope are identified; the announced vertical merits direct procurement follow-up.", "score": null, "href": "/providers/ovhcloud-ai-endpoints/#military-contracts" }, "israel-palestine": { "summary": "We did not find OVH Groupe SA among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "OVH Groupe SA is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/providers/ovhcloud-ai-endpoints/#israel-palestine" } }, "sources": [ { "url": "https://www.ovhcloud.com/en-gb/public-cloud/ai-endpoints/", "title": "OVHcloud AI Endpoints", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://corporate.ovhcloud.com/sites/default/files/external_files/2026-04-09-ovhcloud-h1-fy26-pr-eng_vdef_0.pdf", "title": "OVHcloud H1 FY2026 results and Defence vertical plans", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://corporate.ovhcloud.com/sites/default/files/2025-11/sustainability_statement.pdf", "title": "OVHcloud 2025 Sustainability Statement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://corporate.ovhcloud.com/asia/sustainability/environment/", "title": "Managing our environmental impacts at OVHcloud", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://corporate.ovhcloud.com/en-sg/newsroom/news/environmental-impact-tracker/", "title": "OVHcloud Environmental Impact Tracker announcement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://corporate.ovhcloud.com/en-gb/sustainability/", "title": "OVHcloud Sustainability Overview", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/ovhcloud-ai-endpoints/", "overall": { "score": 6, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 9 } }, { "id": "nebius-token-factory", "name": "Nebius Token Factory", "category": "providers", "organization": "Nebius Group N.V.", "website": "https://nebius.com/ai-studio", "summary": "Cloud GPU and inference platform serving selected model families; model hosting and original model/data rights are separate.", "scope": "Nebius is infrastructure/inference provider and may make models from other developers available.", "dimensions": { "energy": { "summary": "Nebius documents Token Factory inference regions and infrastructure options, but the reviewed service pages provide no inference-linked power, emissions, or water measure.", "sourceUrls": [ "https://nebius.com/services/token-factory/inference-service", "https://nebius.com/services/token-factory" ], "confidence": "low", "unknowns": "Request-level electricity, emissions, and water use remain unknown.", "suggestedScore": null, "scoreReason": "No Token Factory region- or request-level energy evidence was established.", "assessmentType": "argument", "argument": "The service page helps identify where inference can run but does not attach energy or water data to a region, model, or request. Request the selected deployment’s measured or method-based footprint before comparing it with other hosts.", "score": null, "href": "/providers/nebius-token-factory/#energy" }, "training-rights": { "summary": "Nebius Token Factory terms incorporate the selected model provider’s EULA and model-specific rules; Nebius’s host/privacy materials do not independently verify training-corpus rights. Model access and customer-input handling therefore must not be treated as corpus provenance.", "sourceUrls": [ "https://nebius.com/trust-center/iso-27018", "https://docs.tokenfactory.nebius.com/legal/terms-of-service" ], "confidence": "high", "unknowns": "Both Nebius-hosted customer data use and selected model corpus rights require product/model-specific terms.", "suggestedScore": null, "scoreReason": "Reviewed cloud/privacy sources do not establish model-corpus rights.", "assessmentType": "conditional", "argument": "Nebius Token Factory terms incorporate the selected model provider’s EULA and model-specific rules; Nebius’s host/privacy materials do not independently verify training-corpus rights. Model access and customer-input handling therefore must not be treated as corpus provenance. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/nebius-token-factory/#training-rights" }, "privacy": { "summary": "Nebius says Token Factory offers a zero-retention mode in which requests and outputs are not stored or reused for training; enterprise pages also describe isolated capacity, EU/US regional deployment, and optional custom DPAs. This is an available configuration, not evidence that every default serverless request uses it; the general privacy policy does not cover processor data handled under a separate DPA.", "sourceUrls": [ "https://nebius.com/trust-center/iso-27018", "https://nebius.com/services/token-factory/enterprise-grade-inference" ], "confidence": "medium", "unknowns": "The public product material does not say the zero-retention mode is the default for all endpoints; verify plan, region, subprocessors, batch/stateful features, and the applicable DPA.", "suggestedScore": 7, "scoreReason": "A documented zero-retention mode and regional/dedicated controls are meaningful, but the strongest protections are configuration- and contract-dependent and the public controller privacy notice excludes processor data.", "assessmentType": "conditional", "argument": "Nebius provides a credible privacy path through explicit zero-retention inference, regional placement, and optional DPA terms, but the protections depend on endpoint and contract. Confirm the plan, region, subprocessor list, and any batch or stateful-feature retention in the service order before sending sensitive content.", "score": 7, "href": "/providers/nebius-token-factory/#privacy" }, "openness": { "summary": "Nebius provides a catalog of upstream open models and lets customers upload and serve their own fine-tuned or LoRA models, but Token Factory is a managed inference stack. Model licenses and the degree to which weights can be relocated are model-specific; the service docs do not establish that the runtime itself is open.", "sourceUrls": [ "https://nebius.com/services/token-factory", "https://nebius.com/services/token-factory/inference-service" ], "confidence": "medium", "unknowns": "Open model availability and license depend on selected model; model catalogue and per-model cards need review.", "suggestedScore": null, "scoreReason": "Token Factory exposes a hosted catalog of selected open models and model/API documentation, but is not an open serving stack and does not disclose all model training data.", "assessmentType": "conditional", "argument": "Nebius documents customer model uploads and access to open models, but those options do not make its proprietary serving stack inspectable or portable. Check the chosen model license and portability separately, and leave the provider-stack score unrated until runtime source or independent deployment instructions are available.", "score": null, "href": "/providers/nebius-token-factory/#openness" }, "labour": { "summary": "Nebius’s group Code of Business Ethics and supplier-policy pages describe expected conduct, but do not publish worker-level outcomes or audit coverage for Token Factory’s model-data and infrastructure suppliers.", "sourceUrls": [ "https://assets.nebius.com/assets/1bc2c208-d045-4bf0-a40c-8792c53cbb07/66d9df230cbdd48d644f0e4f_Code%20of%20Business%20Ethics%20and%20Conduct.pdf?cache-buster=2025-08-26T17%3A21%3A17.133Z", "https://nebius.com/corporate-policies" ], "confidence": "low", "unknowns": "Conditions for contractors, annotators, and model evaluators remain unknown.", "suggestedScore": null, "scoreReason": "Unrated: the reviewed sources do not establish worker-level outcomes.", "assessmentType": "argument", "argument": "These policies establish standards rather than evidence that workers in the service’s supply chain experience them. Ask for supplier coverage, worker grievance and remediation data, and independent audit results specific to the Token Factory chain before treating the commitments as verified safeguards.", "score": null, "href": "/providers/nebius-token-factory/#labour" }, "military-contracts": { "summary": "Nebius’s Clarifai announcement expressly says the licensed inference IP excludes Clarifai’s U.S. government and defense commercial arrangements; it does not establish whether Nebius has other defense contracts or deployments.", "sourceUrls": [ "https://nebius.com/newsroom/nebius-welcomes-clarifai-s-core-team-and-licenses-inference-ip-to-strengthen-nebius-token-factory", "https://nebius.com/services/token-factory" ], "confidence": "low", "unknowns": "No inference about undisclosed customers or contracts is possible from silence.", "suggestedScore": null, "scoreReason": "Unrated: no relevant public defense deployment surfaced in the reviewed sources.", "assessmentType": "argument", "argument": "That exclusion narrows the scope of one transferred asset and should not be read as a company-wide no-defense commitment. Treat Token Factory’s defense relationship as unknown until a direct contract or deployment is documented, and check the model, account, and customer scope.", "score": null, "href": "/providers/nebius-token-factory/#military-contracts" }, "israel-palestine": { "summary": "We did not find Nebius Group N.V. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Nebius Group N.V. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": { "summary": "Nebius says its Israel AI Cloud deployment in Modi’in offers compute for local companies and research institutes. Its current Token Factory subprocessor disclosure lists Nebius Israel Ltd. for local model processing and service operations. These establish infrastructure and business presence, not a military or rights-related use.", "unknowns": "The company disclosures do not identify military customers, surveillance use, or a conflict-related outcome. A local facility or corporate identity is not treated as harm evidence.", "sourceUrls": [ "https://nebius.com/newsroom/nebius-brings-nvidia-blackwell-to-israel-with-one-of-the-country-s-first-ai-infrastructure-deployments", "https://docs.nebius.com/legal/sub-processors_tofa" ] }, "score": null, "href": "/providers/nebius-token-factory/#israel-palestine" } }, "sources": [ { "url": "https://nebius.com/trust-center/iso-27018", "title": "Nebius ISO/IEC 27018", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://nebius.com/newsroom/nebius-brings-nvidia-blackwell-to-israel-with-one-of-the-country-s-first-ai-infrastructure-deployments", "title": "Nebius brings NVIDIA Blackwell to Israel with one of the country’s first AI infrastructure deployments", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.nebius.com/legal/sub-processors_tofa", "title": "Nebius Token Factory Sub-processors", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://nebius.com/services/token-factory/enterprise-grade-inference", "title": "Nebius Token Factory enterprise-grade inference", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.tokenfactory.nebius.com/legal/terms-of-service", "title": "Nebius Token Factory Terms of Service", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://nebius.com/services/token-factory", "title": "Nebius Token Factory", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://nebius.com/services/token-factory/inference-service", "title": "Nebius Token Factory inference service", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://assets.nebius.com/assets/1bc2c208-d045-4bf0-a40c-8792c53cbb07/66d9df230cbdd48d644f0e4f_Code%20of%20Business%20Ethics%20and%20Conduct.pdf?cache-buster=2025-08-26T17%3A21%3A17.133Z", "title": "Nebius Group Code of Business Ethics and Conduct", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://nebius.com/corporate-policies", "title": "Nebius corporate policies", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://nebius.com/newsroom/nebius-welcomes-clarifai-s-core-team-and-licenses-inference-ip-to-strengthen-nebius-token-factory", "title": "Nebius Clarifai inference IP license announcement", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/nebius-token-factory/", "overall": { "score": 7, "rated": 1, "total": 6, "coverage": 0.16666666666666666, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 3 } }, { "id": "cloudflare-workers-ai", "name": "Cloudflare Workers AI", "category": "providers", "organization": "Cloudflare, Inc.", "website": "https://developers.cloudflare.com/workers-ai/", "summary": "Serverless inference on Cloudflare GPUs, with a catalog of third-party open-source models and customer-supplied content. Cloudflare’s data-use commitments apply to Workers AI only; AI Gateway may send content to other providers under their own terms.", "scope": "Managed inference API and model catalog for Workers AI. Excludes training of the listed third-party models and AI Gateway requests processed by another provider.", "dimensions": { "energy": { "summary": "Cloudflare’s 2025 impact report says its Gen 12 servers deliver 84% more requests per kW than Gen 11 and describes replacing 2,700 older servers with 1,661 newer ones. This is company network efficiency evidence; it does not attribute electricity, emissions, or water to Workers AI, so the service footprint remains unrated.", "sourceUrls": [ "https://cf-assets.www.cloudflare.com/slt3lc6tev37/7koyyovVxIqK8zdG1pqo6O/232476d0cc224d793025546295e68b20/Impact-Report-2025_01282026.pdf" ], "confidence": "low", "unknowns": "Workload distribution and infrastructure energy/water attributable to Workers AI remain undisclosed.", "suggestedScore": null, "scoreReason": "Unrated: the reviewed sources do not provide a defensible service-level footprint.", "assessmentType": "argument", "argument": "Cloudflare’s 2025 impact report says its Gen 12 servers deliver 84% more requests per kW than Gen 11 and describes replacing 2,700 older servers with 1,661 newer ones. This is company network efficiency evidence; it does not attribute electricity, emissions, or water to Workers AI, so the service footprint remains unrated. For a deployment, request or export the selected service and region’s usage-linked footprint when available; do not treat company-wide renewable matching as per-request impact.", "score": null, "href": "/providers/cloudflare-workers-ai/#energy" }, "training-rights": { "summary": "Cloudflare says it does not train its own LLMs or Workers AI customer content, and that the catalog contains third-party pretrained models subject to their licenses. Corpus rights therefore remain model-specific.", "sourceUrls": [ "https://developers.cloudflare.com/workers-ai/platform/data-usage/", "https://developers.cloudflare.com/workers-ai/", "https://www.cloudflare.com/trust-hub/responsible-ai/" ], "confidence": "high", "unknowns": "The service pages do not establish provenance, consent, or licensing for every upstream model training corpus; review each model’s terms separately.", "suggestedScore": null, "scoreReason": "Unrated: customer-content protections do not establish training-corpus rights for third-party models.", "assessmentType": "conditional", "argument": "Cloudflare says it does not train its own LLMs or Workers AI customer content, and that the catalog contains third-party pretrained models subject to their licenses. Corpus rights therefore remain model-specific. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/cloudflare-workers-ai/#training-rights" }, "privacy": { "summary": "For standard stateless inference, Cloudflare says it retains prompt and output content only as long as needed to process and return the response and does not train on customer content. Stateful storage, user-enabled logging/caching, batch, files/adapters, legal obligations, and third-party routing are exceptions.", "sourceUrls": [ "https://developers.cloudflare.com/workers-ai/platform/data-usage/" ], "confidence": "high", "unknowns": "The protection depends on feature and configuration. Operational metadata may be retained; AI Gateway requests sent to other providers follow those providers’ practices.", "suggestedScore": 8, "scoreReason": "The published commitment is narrowly scoped to Workers AI stateless inference and excludes training; feature storage, metadata, legal exceptions, and third-party routing are clearly identified.", "assessmentType": "evidence", "argument": "The published commitment is narrowly scoped to Workers AI stateless inference and excludes training; feature storage, metadata, legal exceptions, and third-party routing are clearly identified. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.", "score": 8, "href": "/providers/cloudflare-workers-ai/#privacy" }, "openness": { "summary": "Workers AI offers 50+ third-party open-source models and allows custom model requests, but Cloudflare operates the serverless GPU and serving layer. Open weights are an upstream model property, not an open or portable Workers AI stack.", "sourceUrls": [ "https://developers.cloudflare.com/workers-ai/", "https://developers.cloudflare.com/workers-ai/models/" ], "confidence": "high", "unknowns": "The reviewed sources do not publish the serving runtime or make its infrastructure reproducible outside Cloudflare; each model’s license and portability differ.", "suggestedScore": 4, "scoreReason": "The model catalog includes inspectable/open upstream choices, but the hosted inference stack remains managed and is not itself open or portable.", "assessmentType": "conditional", "argument": "The model catalog includes inspectable/open upstream choices, but the hosted inference stack remains managed and is not itself open or portable. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 4, "href": "/providers/cloudflare-workers-ai/#openness" }, "labour": { "summary": "Cloudflare’s impact report discusses company and supplier responsibility at a broad level, but the reviewed materials do not disclose worker-level outcomes for Workers AI model data, moderation, or data-center supply chains.", "sourceUrls": [ "https://cf-assets.www.cloudflare.com/slt3lc6tev37/7koyyovVxIqK8zdG1pqo6O/232476d0cc224d793025546295e68b20/Impact-Report-2025_01282026.pdf", "https://developers.cloudflare.com/workers-ai/" ], "confidence": "low", "unknowns": "No relevant labor audit, annotation pay disclosure, or worker-outcome report was identified in the reviewed sources.", "suggestedScore": null, "scoreReason": "Unrated: no provider- and service-relevant worker outcome evidence was established.", "assessmentType": "argument", "argument": "Corporate responsibility reporting is not a service-specific audit of the people whose work supports Workers AI. Ask for supplier and contractor coverage, worker grievance channels, and remediation outcomes linked to the actual infrastructure and models used.", "score": null, "href": "/providers/cloudflare-workers-ai/#labour" }, "military-contracts": { "summary": "Cloudflare advertises defense and intelligence services, but its defense page does not establish that Workers AI is included in a named military contract or deployment.", "sourceUrls": [ "https://www.cloudflare.com/defense/", "https://developers.cloudflare.com/workers-ai/" ], "confidence": "low", "unknowns": "This is a bounded source review, not evidence that no undisclosed government customer exists.", "suggestedScore": null, "scoreReason": "Unrated: no relevant defense deployment surfaced in the reviewed sources.", "assessmentType": "argument", "argument": "The company’s defense-sector offering is relevant commercial context, but it is not endpoint-level evidence for Workers AI. Keep this product-specific relationship unrated until the award and model/API scope are disclosed; do not transfer a general defense relationship to every Cloudflare product.", "score": null, "href": "/providers/cloudflare-workers-ai/#military-contracts" }, "israel-palestine": { "summary": "We did not find Cloudflare, Inc. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Cloudflare, Inc. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/providers/cloudflare-workers-ai/#israel-palestine" } }, "sources": [ { "url": "https://developers.cloudflare.com/workers-ai/", "title": "Cloudflare Workers AI overview", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://developers.cloudflare.com/workers-ai/platform/data-usage/", "title": "Your Data and Workers AI", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://developers.cloudflare.com/workers-ai/models/", "title": "Workers AI model catalog", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.cloudflare.com/trust-hub/responsible-ai/", "title": "Cloudflare’s approach to responsible AI", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://cf-assets.www.cloudflare.com/slt3lc6tev37/7koyyovVxIqK8zdG1pqo6O/232476d0cc224d793025546295e68b20/Impact-Report-2025_01282026.pdf", "title": "Cloudflare Impact Report 2025: Network", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.cloudflare.com/defense/", "title": "Cloudflare for Defense and Intelligence", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://www.cloudflare.com/cloudflare-for-government/", "title": "Cloudflare for Government", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/cloudflare-workers-ai/", "overall": { "score": 6, "rated": 2, "total": 6, "coverage": 0.3333333333333333, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 7 } }, { "id": "lambda-inference", "name": "Lambda Inference API (winding down)", "category": "providers", "organization": "Lambda, Inc.", "website": "https://lambda.ai/inference", "summary": "Lambda’s current product page says the Inference API is winding down and directs users to deploy on Lambda GPU instances. Its API-specific privacy statements come from 2025 staff replies and docs; this record does not imply the API remains generally available.", "scope": "The managed Lambda Inference API only. Excludes the separate Lambda GPU Cloud product and customer-operated model deployments.", "dimensions": { "energy": { "summary": "Lambda’s Inference API page says the service is winding down, while its API documentation describes endpoint use without publishing power, emissions, or water data for that route.", "sourceUrls": [ "https://lambda.ai/inference", "https://docs.lambda.ai/public-cloud/lambda-inference-api/" ], "confidence": "low", "unknowns": "The product page does not disclose service-level footprint or workload location.", "suggestedScore": null, "scoreReason": "Unrated: no defensible service-level footprint was found.", "assessmentType": "argument", "argument": "The service lifecycle notice and API docs identify the relevant product boundary, but no footprint can be attributed to its workload or facility. For any remaining or replacement deployment, request the route’s region and usage-linked footprint rather than rely on Lambda’s broader compute claims.", "score": null, "href": "/providers/lambda-inference/#energy" }, "training-rights": { "summary": "Lambda hosted third-party models; the API-specific no-training statement concerns customer prompts and completions, not rights in model training corpora.", "sourceUrls": [ "https://deeptalk.lambda.ai/t/inference-api-privacy/4553" ], "confidence": "medium", "unknowns": "The reviewed sources do not establish corpus provenance or licensing across hosted models.", "suggestedScore": null, "scoreReason": "Unrated: prompt-handling assurances do not establish rights in upstream model training data.", "assessmentType": "conditional", "argument": "Lambda hosted third-party models; the API-specific no-training statement concerns customer prompts and completions, not rights in model training corpora. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.", "score": null, "href": "/providers/lambda-inference/#training-rights" }, "privacy": { "summary": "In 2025 Lambda staff said prompts and completions were encrypted, not retained after inference, and not used for training or other purposes. The current page says the API is winding down; these API-specific statements were not found in Lambda’s general privacy policy, so their current operational scope is uncertain.", "sourceUrls": [ "https://lambda.ai/inference", "https://deeptalk.lambda.ai/t/inference-api-privacy/4553" ], "confidence": "medium", "unknowns": "The API lifecycle and current endpoint availability are unclear, and the cited statement was made in a staff forum rather than a current product contract or privacy policy.", "suggestedScore": null, "scoreReason": "No current API privacy commitment was verified: the no-retention statement is a 2025 staff-forum disclosure, the current product page says the API is winding down, and current endpoint terms were not found.", "assessmentType": "argument", "argument": "No score is assigned because the only API-specific no-retention/no-training statement located is a 2025 staff-forum disclosure, not a current product commitment. Lambda now says the API is winding down, and the reviewed terms do not establish whether that historical promise still applies to an available endpoint. Verify current service status and contractual privacy terms before sending sensitive data.", "score": null, "href": "/providers/lambda-inference/#privacy" }, "openness": { "summary": "Lambda’s API exposed hosted open-source models behind a managed endpoint. Lambda now says the API is winding down and points users to its separate GPU-instance product; users can self-host eligible weights there, but the endpoint/runtime itself is not open.", "sourceUrls": [ "https://lambda.ai/inference", "https://docs.lambda.ai/public-cloud/lambda-inference-api/" ], "confidence": "high", "unknowns": "The current end-of-life timetable and model-by-model availability are not stated on the product page.", "suggestedScore": 4, "scoreReason": "Open upstream weights can move to customer-run GPU deployments, but the managed API itself is closed and winding down.", "assessmentType": "conditional", "argument": "Open upstream weights can move to customer-run GPU deployments, but the managed API itself is closed and winding down. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.", "score": 4, "href": "/providers/lambda-inference/#openness" }, "labour": { "summary": "Lambda’s careers page describes its internal compute-infrastructure workforce and hiring, but does not report worker outcomes for data labeling, contractors, or the suppliers supporting this winding-down inference API.", "sourceUrls": [ "https://lambda.ai/careers", "https://lambda.ai/inference" ], "confidence": "low", "unknowns": "No relevant annotation pay, worker conditions, or independent labor audit was identified.", "suggestedScore": null, "scoreReason": "Unrated: no service-relevant worker outcome evidence was established.", "assessmentType": "argument", "argument": "Company recruiting material says who Lambda hires, not how external workers in the service chain are treated. Since the API is winding down, confirm the actual serving provider and request vendor-level labor and audit evidence for that route before relying on Lambda-level claims.", "score": null, "href": "/providers/lambda-inference/#labour" }, "military-contracts": { "summary": "Lambda’s government CRADA datasheet describes agency and defense-partner infrastructure work, but does not tie those relationships to the winding-down Inference API.", "sourceUrls": [ "https://lambda.ai/hubfs/Datasheet%20-%20CRADA.pdf", "https://lambda.ai/inference" ], "confidence": "low", "unknowns": "This bounded review cannot rule out undisclosed customers or use of separate Lambda infrastructure.", "suggestedScore": null, "scoreReason": "Unrated: no relevant defense deployment surfaced in the reviewed sources.", "assessmentType": "argument", "argument": "The company’s defense-related infrastructure relationships are relevant at the organization level, while this API’s customer and endpoint scope remains unestablished. Keep the API relationship unrated until a contract names the model-serving service and deployment; verify the replacement route if the API has been retired.", "score": null, "href": "/providers/lambda-inference/#military-contracts" }, "israel-palestine": { "summary": "We did not find Lambda, Inc. among the named targets in the two official BNC campaign lists checked on 2026-10-23. This is a limited list check, not BDS approval.", "argument": "Lambda, Inc. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.", "assessmentType": "evidence", "confidence": "medium", "unknowns": "This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.", "suggestedScore": null, "scoreReason": "BDS campaign status is a factual category, not a 1–10 ethical rating.", "sourceUrls": [ "https://bdsmovement.net/Guide-to-BDS-Boycott", "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide" ], "bdsStatus": "not-listed", "statusLabel": "No named target found", "reviewedAt": "2026-10-23", "assessmentScope": "Named campaign lists", "bdsMatch": null, "upstreamExposure": null, "businessContext": null, "score": null, "href": "/providers/lambda-inference/#israel-palestine" } }, "sources": [ { "url": "https://lambda.ai/inference", "title": "Run inference your way — Lambda Inference API lifecycle notice", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://docs.lambda.ai/public-cloud/lambda-inference-api/", "title": "Using the Lambda Inference API", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://deeptalk.lambda.ai/t/inference-api-privacy/4553", "title": "Inference API privacy statements from Lambda staff", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://lambda.ai/legal/terms-of-service", "title": "Lambda Terms of Service and Cloud AUP", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://lambda.ai/hubfs/Datasheet%20-%20CRADA.pdf", "title": "Lambda government and defense CRADA datasheet", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://lambda.ai/careers", "title": "Careers at Lambda", "kind": "primary", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/Guide-to-BDS-Boycott", "title": "BNC: live guide to consumer boycott and pressure targets", "kind": "campaign", "checkedAt": "2026-10-23" }, { "url": "https://bdsmovement.net/no-tech-oppression-apartheid-or-genocide", "title": "BNC: No Tech for Oppression, Apartheid or Genocide campaign", "kind": "campaign", "checkedAt": "2026-10-23" } ], "href": "/providers/lambda-inference/", "overall": { "score": 4, "rated": 1, "total": 6, "coverage": 0.16666666666666666, "partial": true, "color": "yellow", "method": "Equal-weight mean of available numerical ratings; unrated dimensions and BDS status are excluded.", "rank": 19 } } ], "logos": { "harnesses/claude-code": { "path": "/assets/logos/claude-color.svg", "label": "Claude Code", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/claude-color.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/cursor": { "path": "/assets/logos/cursor.svg", "label": "Cursor", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/cursor.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/gemini-cli": { "path": "/assets/logos/gemini-color.svg", "label": "Gemini CLI", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/gemini-color.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/codex-cli": { "path": "/assets/logos/codex-color.svg", "label": "OpenAI Codex CLI", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/codex-color.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/roo-code": { "path": "/assets/logos/roocode.svg", "label": "Roo Code (archived)", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/roocode.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/github-copilot": { "path": "/assets/logos/copilot-color.svg", "label": "GitHub Copilot", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/copilot-color.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/windsurf": { "path": "/assets/logos/windsurf.svg", "label": "Windsurf", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/windsurf.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/goose": { "path": "/assets/logos/goose.svg", "label": "Goose", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/goose.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/opencode": { "path": "/assets/logos/opencode.svg", "label": "OpenCode", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/opencode.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/openhands": { "path": "/assets/logos/openhands-color.svg", "label": "OpenHands", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/openhands-color.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/kilo-code": { "path": "/assets/logos/kilocode.svg", "label": "Kilo Code", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/kilocode.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/devin": { "path": "/assets/logos/devin-color.svg", "label": "Devin", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/devin-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/gpt-6": { "path": "/assets/logos/openai.svg", "label": "GPT-6 family", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/openai.svg", "license": "MIT; brand trademark remains with owner" }, "models/claude-5": { "path": "/assets/logos/claude-color.svg", "label": "Claude 5 family", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/claude-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/gemini-4-argon": { "path": "/assets/logos/gemini-color.svg", "label": "Gemini 4 Argon", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/gemini-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/grok-4-7": { "path": "/assets/logos/grok.svg", "label": "Grok 4.7", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/grok.svg", "license": "MIT; brand trademark remains with owner" }, "models/llama-4": { "path": "/assets/logos/meta-color.svg", "label": "Llama 4 family", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/meta-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/deepseek-v4-1": { "path": "/assets/logos/deepseek-color.svg", "label": "DeepSeek V4 family", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/deepseek-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/qwen-3-8": { "path": "/assets/logos/qwen-color.svg", "label": "Qwen3.8 family", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/qwen-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/mistral-large-4": { "path": "/assets/logos/mistral-color.svg", "label": "Mistral Large 4", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/mistral-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/phi-4": { "path": "/assets/logos/microsoft-color.svg", "label": "Phi-4 family", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/microsoft-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/command-a-plus": { "path": "/assets/logos/cohere-color.svg", "label": "Command A+", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/cohere-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/amazon-nova-2": { "path": "/assets/logos/nova-color.svg", "label": "Amazon Nova 2 family", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/nova-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/nemotron-3-5": { "path": "/assets/logos/nvidia-color.svg", "label": "Nemotron 3.5 family", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/nvidia-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/olmo-3-7b-instruct": { "path": "/assets/logos/ai2-color.svg", "label": "Olmo 3 7B Instruct", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/ai2-color.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/replit-agent": { "path": "/assets/logos/replit-color.svg", "label": "Replit Agent", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/replit-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/kimi-k3": { "path": "/assets/logos/kimi-color.svg", "label": "Kimi K3", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/kimi-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/glm-5-3": { "path": "/assets/logos/zai.svg", "label": "GLM-5.3 family", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/zai.svg", "license": "MIT; brand trademark remains with owner" }, "models/minimax-m3-1-flash": { "path": "/assets/logos/minimax-color.svg", "label": "MiniMax M3.1 Flash", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/minimax-color.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/zed": { "path": "/assets/logos/zed.jpg", "label": "Zed", "source": "https://avatars.githubusercontent.com/u/79345384?s=64&v=4", "license": "Brand mark for product identification; no affiliation" }, "providers/openai-api": { "path": "/assets/logos/openai.svg", "label": "OpenAI API", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/openai.svg", "license": "MIT; brand trademark remains with owner" }, "providers/anthropic-api": { "path": "/assets/logos/anthropic.svg", "label": "Anthropic API", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/anthropic.svg", "license": "MIT; brand trademark remains with owner" }, "providers/google-vertex-ai": { "path": "/assets/logos/googlecloud-color.svg", "label": "Google Vertex AI", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/googlecloud-color.svg", "license": "MIT; brand trademark remains with owner" }, "providers/aws-bedrock": { "path": "/assets/logos/bedrock-color.svg", "label": "AWS Bedrock", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/bedrock-color.svg", "license": "MIT; brand trademark remains with owner" }, "providers/azure-ai-foundry": { "path": "/assets/logos/azure-color.svg", "label": "Azure AI Foundry", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/azure-color.svg", "license": "MIT; brand trademark remains with owner" }, "providers/together-ai": { "path": "/assets/logos/together-color.svg", "label": "Together AI", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/together-color.svg", "license": "MIT; brand trademark remains with owner" }, "providers/fireworks-ai": { "path": "/assets/logos/fireworks-color.svg", "label": "Fireworks AI", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/fireworks-color.svg", "license": "MIT; brand trademark remains with owner" }, "providers/groq": { "path": "/assets/logos/groq.svg", "label": "GroqCloud", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/groq.svg", "license": "MIT; brand trademark remains with owner" }, "providers/cerebras": { "path": "/assets/logos/cerebras-color.svg", "label": "Cerebras Inference", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/cerebras-color.svg", "license": "MIT; brand trademark remains with owner" }, "providers/huggingface-inference-endpoints": { "path": "/assets/logos/huggingface-color.svg", "label": "Hugging Face Inference Endpoints", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/huggingface-color.svg", "license": "MIT; brand trademark remains with owner" }, "providers/replicate": { "path": "/assets/logos/replicate.svg", "label": "Replicate", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/replicate.svg", "license": "MIT; brand trademark remains with owner" }, "providers/deepinfra": { "path": "/assets/logos/deepinfra-color.svg", "label": "DeepInfra", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/deepinfra-color.svg", "license": "MIT; brand trademark remains with owner" }, "providers/openrouter": { "path": "/assets/logos/openrouter-color.svg", "label": "OpenRouter", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/openrouter-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/apertus-1-5-8b": { "path": "/assets/logos/apertus-1-5-8b.ico", "label": "Apertus 1.5 8B", "source": "https://www.apertus-ai.org/favicon/favicon.ico", "license": "Brand mark for product identification; no affiliation" }, "models/apertus-8b-2509": { "path": "/assets/logos/apertus-8b-2509.ico", "label": "Apertus-8B-2509", "source": "https://www.apertus-ai.org/favicon/favicon.ico", "license": "Brand mark for product identification; no affiliation" }, "harnesses/continue": { "path": "/assets/logos/continue.png", "label": "Continue", "source": "https://avatars.githubusercontent.com/u/127876214?s=64&v=4", "license": "Brand mark for product identification; no affiliation" }, "providers/nebius-token-factory": { "path": "/assets/logos/nebius.svg", "label": "Nebius Token Factory", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/nebius.svg", "license": "MIT; brand trademark remains with owner" }, "harnesses/aider": { "path": "/assets/logos/aider.png", "label": "Aider", "source": "https://avatars.githubusercontent.com/u/172139148?s=64&v=4", "license": "Brand mark for product identification; no affiliation" }, "providers/ovhcloud-ai-endpoints": { "path": "/assets/logos/ovhcloud-ai-endpoints.ico", "label": "OVHcloud AI Endpoints", "source": "https://www.ovhcloud.com/favicon.ico", "license": "Brand mark for product identification; no affiliation" }, "models/reflection-beam": { "path": "/assets/logos/reflection-beam.ico", "label": "Reflection Beam", "source": "https://reflection.ai/favicon.ico?favicon.3aeq0adjuwkbp.ico", "license": "Brand mark for product identification; no affiliation" }, "harnesses/cline": { "path": "/assets/logos/cline.svg", "label": "Cline", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/cline.svg", "license": "MIT; brand trademark remains with owner" }, "models/pleias-pico": { "path": "/assets/logos/pleias.png", "label": "Pleias", "source": "https://www.pleias.fr/favicon-32x32.png", "license": "Brand mark for product identification; no affiliation" }, "providers/scaleway": { "path": "/assets/logos/scaleway-generative-apis.svg", "label": "Scaleway Generative APIs", "source": "https://www.scaleway.com/favicon/website/favicon.svg", "license": "Brand mark for product identification; no affiliation" }, "providers/infomaniak": { "path": "/assets/logos/infomaniak-ai-services.ico", "label": "Infomaniak AI Services", "source": "https://www.infomaniak.com/favicon.ico", "license": "Brand mark for product identification; no affiliation" }, "providers/cloudflare-workers-ai": { "path": "/assets/logos/cloudflare-color.svg", "label": "Cloudflare", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/cloudflare-color.svg", "license": "MIT; brand trademark remains with owner" }, "providers/lambda-inference": { "path": "/assets/logos/lambda.svg", "label": "Lambda", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/lambda.svg", "license": "MIT; brand trademark remains with owner" }, "models/gemma": { "path": "/assets/logos/gemma-color.svg", "label": "Google Gemma", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/gemma-color.svg", "license": "MIT; brand trademark remains with owner" }, "models/gpt-oss": { "path": "/assets/logos/openai.svg", "label": "OpenAI", "source": "https://unpkg.com/@lobehub/icons-static-svg@1.95.1/icons/openai.svg", "license": "MIT; brand trademark remains with owner" } } }