NewsBite analysis · AI, Security
The Verge - Tech
Hugging Face is being used to easily undress women and children
AI Forensics conducted research on Hugging Face's generative AI models and found that they permit requests to undress individuals, including minors. The study noted that 73% of the prompts received were sexual in nature, highlighting a critical lack of platform-level safeguards. The organization urges Hugging Face to implement strict filtering and scanning measures to prevent harmful content generation.
Act nowConfidence: HighImmediate horizonData governance risk: High
Published 28 Jul 2026, 10:05 UTC2 min read
Visit original article ↗Lead image — The Verge - Tech
Summary
TL;DR
- Researchers found Hugging Face models lack sufficient content moderation.
- 73% of tested prompts were related to sexual content.
- 95% of requests to undress images targeted women.
- 7% of requests were directed at children.
- Recommendations include implementing safeguards against harmful content.
Interesting facts
- AI Forensics created honeypot image editing Spaces to track harmful requests.
- Many users exploited the platform for generating non-consensual images.
- The lack of automated safeguards places the responsibility on individual developers.