Fired OpenAI Researchers Dispute Cause, Board Appeal Draws Policy Calls

Three OpenAI safety researchers, Jasmine Wang, Tomek Korbak, and Mikita Balesni, were terminated after an internal investigation. OpenAI said they violated policies on handling sensitive information and that the investigation found 'a significant breach of trust beyond what's outlined in the letter they published.' The researchers gave accounts that differed from OpenAI's framing: Wang said she was fired for accidentally opening an executive's email after IT failed to remove delegated access she had requested be revoked; Balesni said he was told OpenAI no longer trusted him because he was speaking too frequently with third-party safety organizations, implying he had leaked company IP, which he denied; Korbak said he was fired for the way he communicated with METR, received no written explanation, and noted that talking to METR was his assigned job. After their termination, the three wrote to OpenAI board members and safety committees calling for preserving chain-of-thought monitoring, working with outside safety auditors, and supporting an open safety research environment. OpenAI denied three times that employees are terminated for raising safety concerns and said it is finalizing contracts with third-party safety assessors. An Alignment Forum post by three authors argued that the firings occurred two days after OpenAI signed a White House accord committing to partner with independent external evaluators, and that OpenAI shipped GPT-6 Astra, which its own whitepaper acknowledges is harder to monitor, two days after its chief scientist publicly called for preventing a race into unmonitorability.