OpenAI dismissed three safety researchers, Jasmine Wang, Tomek Korbak, and Mikita Balesni, on or around October 1, 2026. The company stated the terminations were due to the unauthorized sharing of confidential materials with an external safety organization, citing violations of internal policies.
The researchers published an open letter on October 8, 2026, addressed to the OpenAI board and safety committees. They warned that the industry risks losing the ability to monitor the reasoning processes of advanced AI models, specifically through the chain of thought analysis used to detect potential model risks.
The authors of the letter proposed two primary solutions: fostering industry-wide collaboration on model monitoring and engaging independent safety auditors. OpenAI responded by stating it agrees with these recommendations, maintaining that the dismissals were strictly related to breaches of trust.
The incident follows the resignation of David Robinson, who previously served as OpenAI’s safety transparency lead and had expressed concerns regarding the company's internal culture. Similar personnel departures occurred at the firm in 2024, also involving allegations of policy violations during ongoing safety debates.
