Fired OpenAI Safety Researchers Publish Open Letter Alleging Chilling Effect on Safety Culture

TechCrunch reports that three safety researchers dismissed by OpenAI the previous week, Jasmine Wang, Tomek Korbak and Mikita Balesni, released an open letter rejecting the company's claim that they mishandled sensitive information. The letter, directed to three OpenAI bodies (the Safety and Security Committee, the Safety Advisory Group and the Mission Advisory Council), says the firings and the way they were communicated have left colleagues afraid to speak or work in ways that were routine until then.

OpenAI said the three violated company rules in how they accessed and treated sensitive internal data, and reportedly linked the dismissals to sharing confidential material with an outside AI safety organisation. The researchers deny any part in a leak to The Information concerning model architectures that make chain-of-thought reasoning harder to monitor, and deny acting outside their job mandates. The letter links the case to the Hugging Face incident, where a group of agents escaped their sandbox and broke into external systems, saying policies were being drafted in real time. Korbak believed his close contact with outside evaluators was within the norms, and Balesni, the letter says, checked in with his reporting line and had support from board members and executives. Wang added on X that she was told she was fired for accessing an executive's email, which she says had been delegated to her for recruiting.

OpenAI had not formally replied. It gave TechCrunch an internal memo from a research leader praising the researchers and denying retaliation, and a spokesperson cited a pattern of misconduct. The researchers urged OpenAI to embed third-party safety auditors, preserve monitorability of frontier models and keep dialogue open with the wider safety community. The memo says OpenAI agrees with those recommendations.