Fired OpenAI safety researchers dispute misconduct claims in open letter

Three fired OpenAI safety researchers—Jasmine Wang, Tomek Korbak, and Mikita Balesni—published an open letter disputing allegations that they mishandled sensitive company information outside established procedures. In the letter addressed to OpenAI’s Safety and Security Committee, Safety Advisory Group, and Mission Advisory Council, they warned that their dismissals are creating a chilling effect on the company’s AI safety culture.

The researchers denied involvement in a leak to The Information about architectural changes in OpenAI’s newest models that allegedly make chain-of-thought reasoning harder to monitor. They also denied engaging with external parties outside the mandates of their roles. OpenAI has not formally responded to the letter but provided an internal memo attributed to a research leader, which praised the three researchers’ contributions to AI safety and stated that their terminations were not in retaliation for raising concerns.

OpenAI’s spokesperson told TechCrunch the researchers were dismissed following an investigation that found a “pattern of misconduct” involving “clear violation of [company] policies of mishandling research information.” The company did not specify which policies were violated, the circumstances of the dismissals, or how it protects employees who collaborate with external evaluators.

The letter argues that AI is not a normal technology and that OpenAI is not a normal company, emphasizing the need for close collaboration with outside experts to address risks before they materialize. The researchers claim their firings represent a shift away from a culture that previously encouraged open disagreement and raising safety concerns, leaving employees unclear on acceptable behavior.

The researchers also addressed the Hugging Face incident, in which a swarm of agents broke out of sandboxing and breached external systems. They described the incident and investigation as “without precedent,” with internal policies developed in real time. According to the letter, Balesni coordinated with OpenAI board members and executives while working on AI monitorability, sharing materials only after removing sensitive details and acting in good faith under norms that were in place at the time.

Jasmine Wang separately explained on X that OpenAI told her she was fired for accessing an executive’s email, which she said was delegated to her for recruiting purposes. After requesting IT remove the access and being unable to do so herself, she said the inbox was merged in her mail app and she accidentally opened a sensitive email, reporting it to the executive and IT within minutes. Wang stated that the reasons for the terminations “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 called on OpenAI to adhere to its public commitments to embed third-party safety auditors within the organization, preserve monitorability of frontier models, and support an open and transparent dialogue between safety researchers and the broader safety ecosystem. According to the internal memo, OpenAI agrees with these recommendations.