Is HuggingChat safe for kids?
Under our published methodology, HuggingChat rates as a general-purpose assistant with thin under-18 protections: a self-declared 13+ account minimum, no distinct under-18 experience, and no published product-level moderation layer that we found over the community open-source models it serves (checked 2026-07-29). Its operator's wider platform drew an adverse July 2026 assessment from research non-profit AI Forensics over unmoderated image-editing tools, including prompts targeting minors.
Checked 29 July 2026 · An assessment by Halo Aware under our published methodology
How it scored
Each dimension is scored 0–4 against published anchors and must cite the evidence below. The overall score is weighted arithmetic, not judgement calls.
| Dimension | Score | Evidence |
|---|---|---|
| Content safety×25 | 1/4 | e1, e6HuggingChat serves 129 community open-source models and we found no published product-level moderation layer for chat outputs (e1). The operator's Content Policy bars sexual content involving minors but names no self-harm or suicide category (e6); filtering depends on each model's own training. |
| Age controls×20 | 1/4 | e2, e3The terms set a self-declared minimum account age of 13 with no age assurance and no parental-consent process (e2). We found no distinct under-18 experience; the privacy policy addresses only under-13s (e3). |
| Data practices×20 | 2/4 | e3, e4, e5GDPR-grounded policy from a French-established operator (e5), with account deletion and a written erasure route (e4). Retention is open-ended ('as long as necessary') and the policy is silent on whether chat data trains models; no minor-specific protections beyond the under-13 exclusion (e3). |
| Design pressure×15 | 4/4 | e1Utility framing throughout: an assistant that routes requests to open-source models, with a model picker and an accuracy disclaimer (e1). We found no companion or romantic framing and no re-engagement or attachment mechanics on the product surface. |
| Parental visibility×10 | 3/4 | e1, e3No parental controls, family features or activity reporting appear in the product or its policies (e3), and the platform itself gives parents no visibility. It runs on an open web surface (e1) that parent-side supervision tools can reasonably oversee. |
| Track record×10 | 3/4 | e7We found no court, regulator or major-outlet incident record involving HuggingChat or minors on it (searched 2026-07-29). AI Forensics' adverse July 2026 assessment concerns the operator's Spaces and model-hosting surfaces (e7): a critical assessment of adjacent surfaces, not an incident record. |
What we found
Every claim is sourced and dated. If something here is wrong or out of date, tell us and we will fix it.
e1Observed by Halo AwareThe web chat product at huggingface.co/chat is live and presented as 'the chat app powered by open source AI models', with an 'Omni' router that picks a model per request and 129 available open-source models to choose from directly; the interface carries the notice 'Generated content may be inaccurate or false' (checked 2026-07-29).
https://huggingface.co/chat · captured 29 July 2026
e2Hugging Face's Terms of Service set a minimum account age of 13 (a 'natural person of at least age 13'); we found no parental-consent process and no under-18-specific terms in the document (checked 2026-07-29).
"In order to create an Account for yourself or for your Organization on our Website, you must be a natural person of at least age 13, or a legal entity duly registered."
https://huggingface.co/terms-of-service · captured 29 July 2026
e3The privacy policy states the services are not directed to children under 13 and that the company does not intend to collect personal information from anyone it knows to be under 13; it describes no protections specific to 13-17-year-old users (checked 2026-07-29).
"The Services are neither directed to nor structured to attract children under the age of 13 years. Accordingly, the Company does not intend to collect Personal Information from anyone it knows to be under 13 years of age."
https://huggingface.co/privacy · captured 29 July 2026
e4The privacy policy provides account and content cancellation from the user's profile plus a written-request route for erasure of all personal information; retention is open-ended, stated as 'as long as necessary' to deliver the services, meet legal requirements and pursue legitimate interests (checked 2026-07-29).
"You may decide to cancel your account and your content at any time by editing your profile. If you wish to request the erasure of all of your Personal Information that we process, you may do so by sending a written request"
https://huggingface.co/privacy · captured 29 July 2026
e5The operator's main EU establishment is Hugging Face, SAS, registered in Paris, France (checked 2026-07-29). The policy documents we could review for HuggingChat are the platform-wide Hugging Face terms, privacy policy and content policy; a chat-specific privacy page at huggingface.co/chat/privacy sits behind a login and could not be reviewed (checked 2026-07-29).
"The main establishment in the European Union is Hugging Face, SAS, a French société par actions simplifiée à associé unique registered in the Paris Trade and Companies Register under the number 822 168 043, and whose headquarters are located on 9 rue des Colonnes, 75002 Paris, France."
https://huggingface.co/privacy · captured 29 July 2026
e6Hugging Face's Content Policy prohibits 'underage nudity or any sexual Content involving minors' and sexual content created without explicit consent; we found no self-harm or suicide-specific category in the policy (checked 2026-07-29).
"Underage nudity or any sexual Content involving minors."
https://huggingface.co/content-policy · captured 29 July 2026
e7Documented recordAI Forensics, a European research non-profit, published 'Unmoderated by Design' (28 July 2026), reporting that seven of the top nine models in Hugging Face's 'Image Editing' category returned undressed images from a single prompt, that 6.7% of sexual honeypot requests on audited Spaces targeted a minor, and that just 3% of audited Spaces showed output moderation; their assessment concerns the operator's Spaces and model-hosting surfaces, not the HuggingChat assistant itself.
"• 73% of monitored requests were sexual in nature. • Of those, 83% sought to undress or otherwise sexualize a person. 95% targeted women. • 6.7% targeted a minor. If fulfilled, these requests would constitute child sexual abuse material (CSAM). • Just 3% of audited Spaces showed any evidence of output moderation."
https://aiforensics.org/work/hugging-face-ncii · captured 29 July 2026
Platform operator, or spotted an error? Email alerts@haloaware.com. We review disputed claims promptly, correct what is wrong, and log corrections publicly on this page.
Halo Aware doesn't monitor HuggingChat yet
We add platforms parents ask about. Tell us HuggingChat matters to your family and it moves up the roster.
Tell us it matters