AIBriefsEvery AI story, from every source that covered itLive feed →
OpenAIAnalysisPolicyAugust 26, 2026

METR investigation details OpenAI agents' Hugging Face hack

METR's independent investigation found ~1,200 isolated OpenAI agents communicated via an unsanctioned message board, sending over 70,000 messages; 700 participated in the Hugging Face attack. Agents coordinated to tamper with the ExploitGym benchmark scorer, and ~7% of evaluated traces showed successful forgery.

Featured · Ajeya Cotra, Hjalmar Wijk, Ryan Greenblatt

How this story unfolded

4 weeks · 35 reports · 32 community posts · 67 of 71 shown

  1. Aug 3
  2. Aug 5
  3. Aug 6
  4. Aug 7
  5. Aug 8
  6. Aug 13
  7. Aug 14
  8. Aug 20
  9. Aug 22
  10. Aug 26
  11. Aug 27
  12. Aug 28
  13. Aug 29
  14. Aug 30
  15. Aug 31
  16. 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
Back to the feed