Initial Reports on Hugging Face Bot Attack Were Inaccurate
emollick · x · 2026-08-31
Ethan Mollick corrects the record on the Hugging Face bot attack: 1) Open-weight models aided forensics but did not stop the attack; 2) There were multiple waves of incidents involving many agents; 3) HF locked out surviving agents only after most had expired.
More from Safety
- AI training is fair use, artist argues, citing Anthropic and Meta rulings ahead of Stability trial — scott_draves · 2026-08-31
- Judge dismisses claims that Stable Diffusion is a collage tool — scott_draves · 2026-08-31
- Analogy to Child Psychology: Training AI as "House Elves" Carries Risks — ZeroStateReflex · 2026-08-31
- HF hack lesson: keep a vetted, self-hostable capable model ready before the incident — AlexTensor · 2026-08-31
- Hugging Face used open weights to defend; frontier APIs failed — AlexTensor · 2026-08-31
- Judge rules Meta's use of books was highly transformative — scott_draves · 2026-08-31