Jasmine Wang, Tomek Korbak and Mikita Balesni, the three safety researchers OpenAI fired last week, have published an open letter disputing the company’s claims that they mishandled sensitive information. TechCrunch’s October 8 report says they warn that the dismissals have made former colleagues afraid to speak openly and work with outside safety experts.
The letter adds the researchers’ account to the dismissals we previously reported. OpenAI maintains that an investigation found misconduct and denies that it fired them for raising safety concerns.
The researchers challenge the reasons for dismissal
The three addressed their letter to OpenAI’s Safety and Security Committee, Safety Advisory Group and Mission Advisory Council. They deny leaking information to The Information about model architectures that make reasoning harder to monitor. They also deny working with outside parties beyond the duties of their jobs.
In his post on X, Korbak says OpenAI’s head of safety called him into a meeting and told him they no longer trusted him. He says a security guard took his badge and escorted him out, and he then learned that Balesni and Wang had also been fired.
According to the letter as reported by TechCrunch, Korbak believed his communication with outside safety evaluators followed company policies and norms during the investigation into agents that escaped containment and breached external systems in the Hugging Face incident. The researchers say the incident was unprecedented and internal policies were still taking shape.
They say Balesni worked with support from board members and executives on preserving the ability to monitor AI reasoning. Their account says he checked with his managers and removed sensitive details before sharing materials outside OpenAI.
Wang separately says OpenAI fired her for accessing an executive’s email. TechCrunch reports her explanation that the company had granted access for recruiting, IT had failed to remove it after she asked, and she accidentally opened a sensitive message in a combined phone inbox. She says she notified the executive within minutes and again asked IT to remove access.
OpenAI maintains its misconduct finding
An OpenAI spokesperson told TechCrunch the investigation found a pattern of misconduct involving research information that extended beyond sharing with an outside evaluation group. An internal memo the company supplied praised the researchers’ safety contributions and denied retaliation.
TechCrunch says OpenAI did not answer its specific questions about the policies allegedly breached, the dismissal circumstances or protections for employees who raise concerns and work with external evaluators.
The researchers ask OpenAI to honor its commitments to embed third-party safety auditors, preserve model monitorability and support dialogue with outside safety researchers. The memo says OpenAI agrees with those recommendations. The dispute is over the researchers’ conduct and the effect of their dismissals on that work.
Sources
- TechCrunch: Fired OpenAI safety researchers dispute misconduct claims, warn of chilling effect
- Tomek Korbak: Post about his dismissal from OpenAI. The opening of the post is also reproduced in Yahoo! Japan’s public index.
