OpenAI defends firing three safety researchers, citing trust breach
OpenAI has defended its decision to dismiss three safety researchers, saying an internal investigation found a serious breach of trust and violations of sensitive-information policies. The company denied that the firings were retaliation for the researchers' public safety concerns. The researchers had published an open letter asking OpenAI to be more transparent about the dismissals.
OpenAI said its decision to dismiss Jasmine Wang, Tomek Korbak, and Mikita Balesni followed an internal review. The company described the findings as a major trust violation and said the three had broken explicit rules for protecting sensitive material. It also rejected claims that their public safety advocacy caused the terminations.
The researchers had asked OpenAI for more openness in an open letter published a day earlier. They argued their conduct matched the organization’s stated mission and the norms in place at the time. OpenAI countered that its inquiry found additional policy breaches beyond those described in the letter, without specifying them.
The dispute may deepen scrutiny of how AI labs balance secrecy with internal safety dissent. Employees who raise concerns could feel more exposed, while companies may face pressure to clarify policies for handling sensitive information and investigations. Researchers, policymakers, and the public could also question whether safety warnings are being heard. If trust erodes, recruitment and collaboration in frontier AI may suffer, and calls for clearer external oversight could grow. However, the episode’s actual consequences remain uncertain.