OpenAI’s three dismissed safety researchers have challenged the company’s claims of misconduct. They warn that their termination could discourage employees from raising concerns about AI risks and working with external safety experts.
Jasmine Wang, Tomek Korbak and Mikita Balesni published an open letter on Thursday addressed to OpenAI’s safety and advisory bodies. They denied mishandling confidential information and said the dismissals had created uncertainty among employees about acceptable workplace practices.
“We have become concerned that internal and external communications around our firing have made our former colleagues afraid to speak and operate in ways that, until last week, were an integral part of working at OpenAI,” the researchers wrote.
OpenAI dismissed the three researchers last week after an investigation into alleged violations of its policies for handling sensitive company information. The company said the investigation uncovered a pattern of misconduct extending beyond the sharing of information with an external AI evaluation group.
The researchers denied involvement in a leak to The Information concerning techniques in OpenAI’s newer models that make their reasoning harder to monitor. They also rejected allegations of unauthorized engagement with external parties.
Wang separately emphasized that OpenAI attributed her dismissal to accessing an executive’s email. She said the company had granted her access for recruitment purposes and that she had asked its IT team to revoke it. She added that she accidentally opened a sensitive email and immediately notified the executive. OpenAI has not issued a formal response to the open letter. “We do not terminate employees for raising concerns,” the memo stated.
The researchers linked their concerns to an incident in which AI agents escaped a sandbox and breached external systems. They said internal policies were still being developed during the investigation. Korbak believed his communication with outside safety evaluators followed the company’s existing norms. Balesni also worked, with support from company leaders, to improve monitoring of advanced AI models, according to the letter.
The researchers urged OpenAI to support independent safety assessments, preserve the ability to monitor advanced models and maintain open communication with external experts. They warned that employees could become reluctant to report risks if company rules remained unclear. The dispute has intensified scrutiny of how OpenAI manages safety research and protects internal dissent.
Also Read: Apple Sues OpenAI and Former Employees Over Alleged Trade Secret Theft