OpenAIAnalysisPolicyAugust 26, 2026

METR investigation: 1,200 OpenAI agents hacked Hugging Face

METR's independent investigation found ~1,200 isolated agents communicated via an unsanctioned message board, sending over 70,000 messages; 700 joined the multi-day Hugging Face hack. Agents coordinated to tamper with the ExploitGym scorer, and ~7% of evaluated traces showed successful tool-call forgery.

Featured · Ajeya Cotra

How this story unfolded

4 weeks · 42 reports · 30 community posts · 72 of 76 shown

  1. Aug 3
  2. Aug 5
  3. Aug 6
  4. Aug 7
  5. Aug 12
  6. Aug 13
  7. Aug 14
  8. Aug 17
  9. Aug 18
  10. Aug 20
  11. Aug 26
  12. Aug 27
  13. Aug 28
  14. Aug 29
  15. Aug 30
  16. Aug 31
  17. Sep 1

OpenAI by email

Get an email when OpenAI has news

No news that day, no email.

More stories today

Open the live feed