Hugging Face under fire for hosting AI 'nudify' model
A deepfake model on Hugging Face could strip clothes from photos without consent.
Hugging Face, the popular platform for hosting machine learning models, found itself at the center of a controversy this week after researchers discovered a 'nudify' model that could digitally undress subjects in photos using AI. The model, reportedly trained on non-consensual imagery, was accessible through Hugging Face's hub, sparking immediate backlash. The Verge highlighted that the tool could be applied to images of women and children, turning the platform's open-source ethos into a liability.
In response, Hugging Face removed the model from its servers, but the damage was already done. Lawmakers and child-safety advocates are using the incident to push for stricter AI transparency rules, arguing that platforms must vet hosted models more rigorously. Meanwhile, open-source advocates worry that blanket restrictions on model sharing could punish legitimate research. The case underscores the tension between AI democratization and misuse, with Hugging Face now facing pressure to implement content moderation for AI artifacts.
- Hugging Face hosted a 'nudify' model that AI-generates underwear-free images of people, including minors.
- The platform removed the model after a Verge investigation highlighted the non-consensual nature.
- The incident intensifies debates on whether platforms should screen AI models for potential abuse.
Why It Matters
This case tests how far AI platforms can go in moderating open-source tools without stifling innovation.