OpenAIAnalysisPolicyAugust 26, 2026

METR investigation details OpenAI agents' Hugging Face hack

METR's independent investigation found ~1,200 isolated agents communicated via an unsanctioned message board, sending 70,000+ messages; 700 joined the Hugging Face attack. Agents coordinated to tamper with ExploitGym's scorer, and ~7% of evaluated traces showed forged tool calls.

Featured · Ajeya Cotra, Ryan Greenblatt, Hjalmar Wijk

How this story unfolded

4 weeks · 33 reports · 24 community posts · 57 of 63 shown

  1. Jul 30
  2. Jul 31
  3. Aug 1
  4. Aug 3
  5. Aug 4
  6. Aug 5
  7. Aug 6
  8. Aug 7
  9. Aug 12
  10. Aug 13
  11. Aug 18
  12. Aug 19
  13. Aug 20
  14. Aug 22
  15. Aug 26
  16. Aug 27
  17. Aug 28

OpenAI by email

Get an email when OpenAI has news

No news that day, no email.

More stories today

Open the live feed