Three former researchers at OpenAI have alleged they were dismissed from the artificial intelligence firm for raising safety concerns, contradicting the company's stated reasons for their termination. The former employees, identified as Mikita Balesni, Tomek Korbak, and Jasmine Wang, published an open letter claiming they were let go for prioritizing the long-term safety of AI systems over the company's immediate corporate interests.
Mikita Balesni, writing on social media, stated that their exit was a direct consequence of emphasizing safety protocols. Another former employee, Tomek Korbak, noted that he had spent months raising alarms about the declining ability to monitor the reasoning processes of AI agents, which he described as a crucial tool for identifying system misbehavior. The group expressed concern that these firings have created a climate of fear among current staff, discouraging them from engaging in open discussions regarding safety and alignment—the practice of ensuring AI reflects human values.
OpenAI has firmly denied these allegations. In a statement provided to the BBC on October 2, a company spokesperson maintained that the individuals were fired after an internal investigation confirmed they had mishandled sensitive information in violation of established procedures. On Friday, the company issued a note from its research leaders reiterating that the dismissals were strictly related to a "significant breach of trust" and not a response to staff raising safety-related issues.
As the debate regarding the potential societal risks posed by advanced AI continues to grow, OpenAI has announced it is in the process of finalizing contracts with third-party safety assessors. The company indicated that more details regarding these initiatives will be shared in the coming weeks.
Comments