Ethical Stack.

HuggingChat

Hosted chat interface that rotates and lets users select open models from the Hugging Face ecosystem.

Score calculation

(3 + 4 + 1 + 7) ÷ 4 = 3.8

4 of six dimensions rated. Unrated dimensions are excluded from the average.

Full scoring method ↗
Setup details

HuggingChat web app and its current model-selection routes; model authors’ own sites and independent inference endpoints are excluded.

What fits your needs

HuggingChat web app and its current model-selection routes; model authors’ own sites and independent inference endpoints are excluded.

Where it fits

Use HuggingChat for a browser-based comparison across current open models, or choose an available model directly.

See the evidence
Tradeoffs to expect

Route and limits

Available models rotate and Omni can route automatically, so repeatability and upstream data handling depend on the selected route.

See the evidence
How to switch

Try it

Open HuggingChat, compare a selected model with Omni on a low-sensitivity prompt, then delete test history if no longer needed.

See the evidence
Practical questions

Before relying on it

Check the active model, current privacy statement, history controls, and any model-specific terms before sensitive use.

See the evidence

Energy practices

3/10↑

Public docs establish the HuggingChat hosted interface and dynamically selected model endpoint route; they don’t show its request-level physical footprint. Users can choose a model, while Omni can route automatically; model choice is a workload control, but the current app publishes no physical energy savings.

See the evidence

This practices and disclosure grade covers HuggingChat hosted interface and dynamically selected model endpoint. Public product/API documentation establishes how this route operates and the listed user controls, but gives no measured per-request energy, emissions or water. A corporate electricity claim is not allocated to this route.

1 + 1 reporting + 0 electricity + 1 controls = 3/10. This grades documented practices and transparency.

Request-level electricity, emissions, water use, serving utilization and any route-specific clean-power matching are not reported for HuggingChat hosted interface and dynamically selected model endpoint.

Reporting1/3
Official documentation describes the HuggingChat hosted interface and dynamically selected model endpoint operating boundary, but does not publish a dated quantitative energy, emissions or water result for this route.
Electricity sourcing0/3
The reviewed sources do not map HuggingChat hosted interface and dynamically selected model endpoint to a site-specific electricity mix or time-matched clean-power supply.
Resource controls1/3
Users can choose a model, while Omni can route automatically; model choice is a workload control, but the current app publishes no physical energy savings.

Water disclosure

Water reporting has not been assessed for this choice.

Training data

—Depends on the model↑

HuggingChat rotates models and Omni can choose a model automatically; identify the selected model and provider before assessing training data provenance and rights.

See the evidence

The app’s policy concerns conversations submitted through HuggingChat, while the model list changes and includes independently published open models. It doesn’t confirm the training data, license, or permission basis for the model selected on a given date.

Selected model version, author and provider, training data source, license scope, training exclusions and creator-remedy options vary by route.

Your data & privacy

—Not enough evidence↑

The HuggingChat privacy statement, last updated in 2024, says conversations are stored for history and not shared with model authors or used for training. The current product also offers Omni automatic routing and selected-model chats, so the applicable inference-provider handling is not established by that statement.

See the evidence

HuggingChat’s published app privacy statement says conversation data is private to the user, stored for history, not shared with model authors or used for training, and deletable. It was last updated April 15, 2024. The current model page describes Omni automatic routing as well as user-selected models; the sources don’t confirm how the current route handles prompts at each inference provider, so a single numeric privacy grade would overstate the evidence.

Current handling of prompts for Omni and each selected model, retention at inference providers, route-specific logs, and whether the 2024 app statement has been updated or supplemented need confirmation.

Openness

4/10↑

The web app provides access to open models, but HuggingChat’s hosted runtime and app infrastructure remain managed by Hugging Face.

See the evidence

HuggingChat gives access to open-source models and a user-selectable model list, while its own app is a hosted Hugging Face service. The presence of open model weights does not make the hosted chat runtime independently deployable.

Open model choices are available; the hosted app’s runtime and request routing are not self-hosted by the user.

Model weights, licenses, hosting providers, and model author data policies vary by selected route.

Worker protections

1/10↑

Hugging Face publishes a community Contributor Covenant with reporting and moderation procedures, but this is a platform-community code, not an employee or contractor labor policy.

See the evidence

The sources reviewed provide product, terms, or corporate information but no qualifying disclosure about conditions for data annotators, evaluators, moderators, or infrastructure workers. This is a disclosure gap within the checked materials, not evidence that workers are mistreated or that no labor policy exists. These findings concern the developer or operator. This product’s use in a particular deployment needs separate evidence.

1 + 0 pay & conditions + 0 worker voice & remedy + 0 evidence & follow-through = 1/10. This grades documented worker safeguards, not a measured rate of mistreatment.

Worker counts, locations, pay, labor protections, worker voice mechanisms, supplier coverage, and independent audit or remediation outcomes were not established in the sources reviewed.

Pay & conditions0/3
Reviewed sources provide no qualifying disclosure of worker conditions; this zero is limited to the checked material and is not an exhaustive claim.
Worker voice & remedy0/3
The cited Contributor Covenant governs platform community participation and reporting, not employees or contractors as workers; no qualifying worker-voice mechanism was established in the reviewed labor sources.
Evidence & follow-through0/3
No independent labor audit results, worker-level corroboration, or remediation outcomes were disclosed in the reviewed material.
Workers rubric ↗

Military ties

7/10Not established↑

No direct military relationship was established for this product or API in the reviewed sources.

See the evidence

The reviewed terms and service documentation do not identify a direct military customer, contract, or deployment for this route. This records the evidence reviewed and does not treat general cloud access, a model license, or downstream use as evidence of a direct contract.

7 base + 0 safeguard adjustment = 7/10 (0 of 9 safeguard points; every 3 adds 1). No military relationship was established in the cited sources. This ranks above known defense ties without claiming a verified clean record.

Customer contracts and downstream uses may not be public; this assessment is limited to the linked service sources.

Use restrictions0/4
The reviewed service sources don’t confirm route-specific military-use restrictions beyond general acceptable-use terms.
Enforcement & oversight0/3
The reviewed sources do not describe independent oversight of military-related deployment of this service.
Relationship disclosure0/2
No service-specific public reporting on military customers or contracts was identified in the reviewed sources.

The reviewed product and service sources don’t show a direct operational military relationship for this specific route.

Military rubric ↗

BDS targets

Not listed↑

Hugging Face was not named in the BNC guide to priority boycott and pressure targets or the No Tech campaign target list reviewed on 24 October 2026. This is limited to those two lists.

See the evidence

The BNC guide and No Tech campaign page reviewed on 24 October 2026 do not name Hugging Face. This status reports only those two named target lists and does not cover every BDS campaign, local group, supplier, investment, or later list change.

Other BDS campaigns and local or partner-led lists may have separate targets; this check does not evaluate broader political or business relationships.

More from Hugging Face

All sources

Side by side