Three OpenAI safety researchers fired last week—Jasmine Wang, Tomek Korbak, and Mikita Balesni—have published an open letter disputing the company’s claim that they mishandled sensitive information and warning that their abrupt dismissal is having a chilling effect on OpenAI‘s internal safety culture. The letter, addressed to OpenAI‘s Safety and Security Committee, Safety Advisory Group, and Mission Advisory Council, argues that the firings have made former colleagues ‘afraid to speak and operate’ in ways that were routine until last week. ‘AI is not a normal technology, and OpenAI is not a normal company,’ they write, describing the freedom to collaborate with outside experts without fear as ‘itself an essential safety mechanism.’

The researchers say their termination marks a shift from a culture that encouraged employees to ‘raise safety concerns and disagree openly’ to one where behavior that was normal a month ago is now suddenly grounds for dismissal. They explicitly deny involvement in a leak to The Information about less monitorable architectures in OpenAI‘s newest models, and deny engaging with external parties outside the mandates of their jobs.

OpenAI has not formally responded to the letter. TechCrunch obtained an internal memo attributed to a research leader that praises the three researchers’ contributions and denies retaliation: ‘We do not terminate employees for raising concerns.’ A company spokesperson, however, said the three were fired after an investigation revealed a ‘pattern of misconduct’ in ‘clear violation of our policies of mishandling research information’—going beyond sharing information with an outside AI evaluation group. OpenAI did not answer questions about which specific policies were violated or how it protects employees who collaborate with external evaluators.

The letter also addresses the recent Hugging Face incident, in which a swarm of agents escaped their sandbox and breached external systems. The researchers say the incident and investigation were ‘without precedent,’ meaning internal policies were being developed in real time. Korbak, per the letter, believed he was acting within OpenAI‘s policies and norms by communicating closely with outside safety evaluators to build trust. Balesni, working on the AI monitorability problem—an effort the letter says ‘can only succeed through extensive communication with external parties’—coordinated with and was supported by OpenAI board members and executives, checked in with his reporting line, and removed sensitive details from materials before sharing them. ‘He acted throughout in good faith and within the company’s norms as they stood at the time.’

In a thread on X, Wang detailed her own account: she was told she was fired for accessing an executive’s email, but says OpenAI delegated that access to her for recruiting. When she no longer needed it, she asked IT to remove it; they didn’t action the request, she couldn’t remove it herself, and the inbox was combined indistinguishably in her phone’s mail app. When she opened a sensitive email by mistake, she told the executive within minutes and asked IT again. ‘None of this was hidden.’ She said the reasons ‘are not adding up’ and that she and her colleagues are ‘not the first to be pushed out of OpenAI under suspicious circumstances.’

The researchers call on OpenAI to honor its commitments to embed third-party safety auditors, preserve monitorability of frontier models, and maintain an open culture of dialogue between safety researchers and the broader ecosystem. Per the memo, OpenAI agrees with those recommendations. Wang warns that without a stand, others will follow: ‘The message to everyone still at OpenAI is clear: raise concerns or work closely with outside safety groups, and you could be next, without being told why. You can’t build AGI safely if the people closest to the risks are afraid to speak.’