Jasmine Wang, Tomek Korbak, and Mikita Balesni, three safety researchers OpenAI fired last week, published an open letter denying the firm’s claims that they mishandled sensitive information outside established procedures and warned their dismissal signals a chilling effect across OpenAI’s culture.
The letter is addressed to OpenAI’s Safety and Security Committee, Safety Advisory Group, and Mission Advisory Council.
OpenAI said they violated policies by “accessing and handling sensitive company information” after allegedly sharing confidential information with a third-party AI safety organization.
The researchers wrote that AI is not a normal technology and that safety work relies on collaboration with outside experts without fear, with clear internal procedures as an essential safety mechanism. They said firing represents a shift from a culture that encouraged raising safety concerns and open disagreement; employees are now unclear when behavior normal a month ago is grounds for dismissal.
They denied involvement in a leak to The Information about less monitorable architectures in newer models that make chain-of-thought reasoning harder to monitor, and denied engaging external parties outside their job mandates.
OpenAI shared an internal memo from a research leader praising their contributions and denying retaliation for raising safety concerns. A spokesperson told TechCrunch the three were fired after an investigation revealed a “pattern of misconduct” in “clear violation of our policies of mishandling research information” beyond sharing with an outside evaluation group. OpenAI did not specify which policies or dismissal circumstances when asked by TechCrunch.
The letter discusses the Hugging Face sandbox breach incident, real-time policy development, Korbak’s communication with outside evaluators, and Balesni’s monitorability work with board and executive support.
Wang said on X that OpenAI told her she was fired for accessing an executive’s email; she said access was delegated for recruiting, IT did not remove it when requested, and she reported opening a sensitive email by mistake within minutes.
They called on OpenAI to embed third-party safety auditors, preserve frontier model monitorability, and support open dialogue with the safety ecosystem. OpenAI’s memo agrees with those recommendations per TechCrunch.
Wang: “You can’t build AGI safely if the people closest to the risks are afraid to speak.”