Ethical Stack

Claude 5.5 family

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.

Score calculation

(4 + 6 + 2 + 4 + 3) ÷ 5 = 3.8

5 of six dimensions rated. Unrated dimensions and BDS status are excluded.

Full scoring method ↗
Setup details

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.

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.

Energy

↑

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.

Reasoning & sources

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.

No comparable per-query or training-run environmental figures found.

Training rights

↑

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.

Reasoning & sources

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.

Training categories are disclosed, but sources and rights are not itemized.

Exact dataset provenance, licensing and contested works are not enumerated in the model report.

Privacy

↑

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.

Reasoning & sources

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.

Consumer training controls and distinct commercial terms are documented, with safety-review and retention exceptions.

Review policy and retention depend on product, settings, and model classification.

Openness

↑

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.

Reasoning & sources

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.

Proprietary hosted model; no public weights or complete training recipe.

No public model weights or complete training data/recipe found.

Worker treatment

↑

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.

Reasoning & sources

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.

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.

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.

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.

Government & military

↑

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.

Reasoning & sources

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.

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.

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.

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.

BDS status

↑

We did not find Anthropic among the named targets in the two official BNC campaign lists. This is a limited list check, not BDS approval.

Reasoning & sources

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.

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.

All sources

Side by side