OpenAI

55

Balesni says he, Wang, and Korbak were fired last week for putting safety ahead of the company. Korbak says a guard took his badge. OpenAI calls it a breach of trust on sensitive information, not the safety talk. The sprint still did not stop.

• x · Oct 8 Balesni: fired last week with two other safety researchers. Letter to leadership. Believes they were fired for prioritizing safety over near-term interests.
• x · Oct 8 Korbak: head of safety said they no longer trust him. A security guard took the badge. Says talking to METR was the job. Nothing in writing.
• web · Oct 9 Reuters: OpenAI post on X. Significant breach of trust beyond the letter. Not for raising safety concerns. Debates happen every day.

Public posts, indexed. Not a work-email tick. Not Glassdoor. OpenAI on Working.