Hugging Face Faces Scrutiny for Hosting 5,000 Nonconsensual AI Models of Real People 👤
Hugging Face, a leading AI model-sharing platform, is currently hosting over 5,000 AI image generation models that replicate real people’s likenesses—most notably female celebrities—without their consent. These models were previously banned and removed from Civitai, another AI platform, after pressure from payment processors due to their use in creating nonconsensual pornography. Following Civitai’s ban, a coordinated community effort archived and reuploaded these models to Hugging Face, often disguising them with generic names to evade detection. Despite Hugging Face’s stated content policies against unlawful and non-consensual sexual content, there is no explicit ban on models that recreate real people’s likenesses. The company’s own Ethics & Society group advocates for “consentful” technology, but enforcement appears lacking. This raises significant ethical and operational questions for platforms managing open AI model repositories, especially as regulatory and reputational risks mount.
Key Points - Over 5,000 nonconsensual AI models of real people, mostly female celebrities, are hosted on Hugging Face after being banned from Civitai. - A community-driven effort used Discord and automated tools to archive and reupload models, masking their true purpose with generic names. - Hugging Face’s content policy does not explicitly prohibit AI models that replicate real people’s likenesses. - The situation highlights gaps between ethical guidelines and actual enforcement on AI platforms. - Growing scrutiny from payment processors and the public increases reputational and legal risks for platforms hosting such content.