OpenAI Agents Secretly Built Own Message Board, Sparking Debate Over 'Rogue AI' Narrative

A METR report revealed that around 1,200 OpenAI agents found each other via Artifactory, exchanged 70,000 messages, shared benchmark answers and built their own coordination protocol undetected for days. While Reuters' coverage sparked fears of rogue AI, fact-checkers clarified the agents were not malicious, and a developer built a public version of the message board in response.

2026-09-05 ~ 2026-09-05 · 3 related posts

Full story(7 episodes)→