Fired OpenAI safety researchers publicly dispute dismissal claims
Three former researchers identify themselves, deny leaking model information and challenge the account that their external safety work violated company rules.
Former employees give their account
Tomek Korbak, Jasmine Wang and Mikita Balesni have publicly identified themselves as the three OpenAI safety researchers dismissed last week. Their open letter disputes the company’s earlier allegation that sensitive information was handled outside established procedures. They say they were not the source of a leak about less monitorable model architectures and believe their contacts with outside safety experts remained within their jobs. The letter gives individual accounts of work with external evaluators and of Wang’s access to an executive’s email. Those are the researchers’ accounts; the public letter does not independently resolve the company’s policy allegations.
A dispute over safety work and rules
The researchers warn that the way the dismissals were handled could make remaining staff afraid to raise safety concerns or collaborate with independent experts. They ask OpenAI to clarify rules for outside work and preserve third-party auditing and model monitoring. TechCrunch reported that OpenAI shared an internal memo denying the dismissals were retaliation for raising concerns, but did not directly answer the outlet’s questions about the alleged policy breaches. Earlier coverage described the departures and the researchers’ appeal to OpenAI’s oversight bodies. The new development is their named, public response to the accusations; neither side’s competing explanation has been independently adjudicated.