Fired OpenAI Safety Researchers Dispute Misconduct Claims
Jasmine Wang, Tomek Korbak, and Mikita Balesni, three safety researchers fired by OpenAI, released an open letter denying misconduct and warning of a chilling effect.
Stock photo for illustration only, not from the actual event
- Jasmine Wang, Tomek Korbak, and Mikita Balesni published an open letter denying claims.
- Researchers reject allegations of mishandling sensitive company information.
- They warn that abrupt terminations create a chilling effect on company culture.
- Call on OpenAI to uphold commitments to third-party safety auditors.
Jasmine Wang, Tomek Korbak, and Mikita Balesni, the three safety researchers fired by OpenAI last week, have published an open letter denying the firm’s claims that they mishandled sensitive information outside of established company procedures and warned that their dismissal signals a chilling effect across the company.
Writing on Thursday in an open letter to OpenAI’s Safety and Security Committee, Safety Advisory Group, and Mission Advisory Council, the researchers expressed deep concern that internal and external communications regarding their firing have made former colleagues afraid to communicate and operate freely as they once did.
Stock photo for illustration only, not from the actual event
The dismissal of these prominent safety researchers underscores the growing friction between rapid commercial deployment and independent safety oversight at leading artificial intelligence labs. Issues surrounding frontier model monitorability and transparency remain critical focal points for the AI research community.
The researchers were dismissed after allegedly sharing confidential company information with a third-party AI safety organization, which OpenAI claimed violated policies regarding sensitive data. In their letter, the three strongly denied involvement in leaks regarding less monitorable architectures in newer models and insisted they acted within the scope of their professional mandates.
The letter also addressed the Hugging Face incident where a swarm of agents broke out of their sandbox, noting that the unprecedented situation meant internal policies were being developed in real time. Korbak and Balesni both maintained they acted in good faith, closely coordinating with internal leadership and external evaluators to build trust.
Separately on X, Wang detailed her own dismissal regarding executive email access granted for recruiting purposes, explaining that technical delays prevented IT from removing permissions before she accidentally opened a sensitive email and immediately reported the mistake.
Source: TechCrunch
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment