OpenAI Fires Three Safety Researchers Over Alleged Leaks to Outside AI Safety Org

OpenAI has fired three researchers from its safety team for allegedly leaking confidential company information to an outside AI safety organization. The departures were first flagged on October 1 by security researcher account @Bayesian00, and on October 2 the Wall Street Journal, citing people familiar with the matter, confirmed the three were terminated, with TechCrunch reporter Max Zeff also covering the story. The situation is still developing, and neither OpenAI nor the safety organization involved has disclosed further details.

Confirmed

• Per WSJ, OpenAI fired 3 safety-team researchers for allegedly passing confidential company information to an external AI safety organization; TechCrunch reporter Max Zeff reported the same.
• An OpenAI spokesperson said an internal investigation found the three failed to handle sensitive information per company procedures, "undermining the trust that is critical to our work" (as quoted via WSJ cited by @rohanpaulai).
• Andrew Curran, citing WSJ, said the three were alignment/safety researchers and were terminated that day, rather than leaving voluntarily.

Not Yet Confirmed

• Where the three are headed: a rumor relayed by dejavucoder suggests they may start an embedded eval/AI safety company, but this is unverified.
• The identity of the third-party safety organization that received the information, and the specific details of what was leaked, remain undisclosed.

Why It Matters

• The episode once again highlights the tension between internal secrecy at frontier labs and external AI safety oversight; personnel changes on safety teams bear directly on model governance and safety capability.
• The "terminated rather than resigned" framing, plus the allegation of leaking confidential material, sets this apart from OpenAI's usual safety-team attrition and has sparked community debate over the boundaries of safety whistleblowing.

2026-10-01 ~ 2026-10-02 · 7 related posts

• [source] Three AI safety researchers reportedly leave OpenAI amid 'lost control' jibes — scaling01 · 2026-10-01
• [source] WSJ: Three OpenAI Alignment/Safety Researchers Were Terminated, Not Just Departed — dejavucoder · 2026-10-02
• OpenAI parts ways with three researchers over allegedly sharing sensitive info with AI safety org — KateClarkTweets · 2026-10-02
• [source] WSJ: OpenAI Fires 3 Safety Researchers Over Leaked Confidential Information — rohanpaul_ai · 2026-10-02
• WSJ: OpenAI Parts Ways With 3 Researchers Over Alleged Leak to AI Safety Group — ResultBackground2450 · 2026-10-02
• OpenAI reportedly parts ways with 3 safety researchers over alleged confidential info leak — Polymarket · 2026-10-02

1 near-duplicate retellings: gwern