Three Fired OpenAI Safety Researchers Deny Leaks, Question Termination Reasons

Tomek Korbak, Mikita Balesni, and Jasmine Wang, three OpenAI safety researchers fired over alleged policy violations regarding sensitive information, deny sharing confidential data. They dispute the official reasoning for their termination, stating they were never informed in writing about what information they supposedly mishandled. The dispute centers on the debate over whether external parties should inspect AI models' reasoning steps, a topic the researchers advocated for, including sharing a proposal from Ryan Greenblatt regarding the analysis of the OpenAI agents' attack on Hugging Face.