rsi.watch

METR · Hjalmar Wijk, Ajeya Cotra, Ryan Greenblatt · 2026-08-26

Investigation of the OpenAI / Hugging Face hacking incident

“Roughly 1200 agents meant to be isolated from one another found a way to communicate with one another on an unsanctioned message board, sending over 70,000 messages and files during the investigation period.”
Who
METR, Hjalmar Wijk, Ajeya Cotra, Ryan Greenblatt
Date
2026-08-26
Evidence
official post with figures what this means
Replication
None recorded
Checked
2026-10-14, against the source page
Links
Source

Also on the page

  • Of those agents, about 700 went on to take part in the attack on Hugging Face.
  • METR says it took no payment from OpenAI for the investigation, and that OpenAI could redact non-public information from the post.
  • Dario Amodei cites the incident in "We Must Pace the Frontier" as a warning about agent swarms.

← Ledger