October 9, 2026

Fired OpenAI Researchers Challenge Dismissals in Open Letter

Three safety researchers dismissed by OpenAI have published an open letter disputing the circumstances of their terminations. They expressed concern that the situation may deter others from discussing AI risks. The researchers, Tomek Korbak, Jasmine Wang, and Mikita Balesni, addressed their letter to OpenAI’s Safety and Security Committee, Safety Advisory Group, and Mission Advisory Council.

In the letter, they claimed that communications regarding their dismissal have made former colleagues fearful of speaking out. Korbak, Wang, and Balesni emphasized the necessity for safety researchers to challenge decisions and collaborate with external experts. They stated, "AI is not a normal technology, and OpenAI is not a normal company."

OpenAI denies that the dismissals were related to their safety advocacy. The company maintains that an investigation revealed violations of policies concerning sensitive information. Newsweek reached out to OpenAI and the researchers for additional comments.

Read the Full Letter

The letter titled "OpenAI cannot make AI safe on its own," warns about the impact of their dismissals on other employees. "We are concerned that communications about our firing have made colleagues afraid to speak up," the researchers wrote. They highlighted OpenAI’s encouragement of open disagreement regarding safety issues but noted a shift in employee clarity.

The researchers further elaborated on their dismissals. Korbak, Wang, and Balesni refuted allegations of leaking information about AI architectures and claimed no unauthorized communications with external organizations. Korbak stated he was told verbally about inappropriate communications with an independent organization but received no specific details.

Balesni discussed perceived misinterpretations by OpenAI. He asserted their firings stemmed from prioritizing safety over the company’s short-term interests. He was reportedly told that OpenAI no longer trusted him due to his interactions with third-party safety organizations but denied ever leaking company intellectual property.

Wang recounted being accused of improper access to an executive’s email. She clarified that OpenAI had granted her access for recruitment purposes and that she reported the issue and sought its resolution.

The researchers made recommendations for OpenAI, urging continued independent safety evaluations and maintaining transparency with external experts. They stressed the importance of monitoring advanced models’ internal reasoning and preserving open communication with safety assessors.

Balesni expressed concern over potential restrictions on OpenAI’s relationship with safety assessors like METR. "I worry OpenAI might use our firing as a reason to sever ties with METR," he stated.

OpenAI issued a statement on October 9 defending its decision. The company cited breaches of trust and policy violations and maintained that dismissals were unrelated to safety concerns being expressed. OpenAI affirmed its commitment to fostering dialogues on safety and research.

The company further emphasized its work with external safety assessors and its intentions to continue collaboration. OpenAI agreed with the researchers’ call for industry-wide cooperation in maintaining model monitorability.

The disagreement between OpenAI and the researchers underscores the balance between confidentiality and external oversight. The researchers’ letter highlights the challenges faced by those aiming to ensure AI safety while operating under organizational constraints.

TAGS: