OpenAI let go of three safety researchers—Jasmine Wang, Tomek Korbak and Mikita Balesni—last week. The company said they broke policy by handling confidential information. The trio responded with an open letter that rejects the accusations and warns that their firing may make other staff reluctant to raise safety concerns or work with outside experts.
What the letter says
In the letter to OpenAI’s Safety and Security Committee and advisory groups, the researchers claim they acted within existing norms while collaborating with third‑party safety groups. They deny leaking any internal details about OpenAI’s newest models, including a story about “less monitorable architectures.” Wang adds that the alleged violation—accessing an executive’s email—was an accidental opening of a delegated inbox that she promptly reported.
OpenAI’s response
OpenAI has not publicly replied to the letter, but it shared an internal memo with TechCrunch that praises the three workers’ contributions and states the dismissals were not retaliation for safety advocacy. The memo says the company “always encourages” raising concerns. The firm did not clarify which specific policies were broken or how it protects employees who engage with external auditors.
Why it matters
If staff at OpenAI feel they could be fired for talking to outside safety groups, they may stop reporting risks, which could slow the detection of dangerous model behavior. The dispute also highlights how unclear internal rules can affect the ability of researchers to collaborate on safety across the industry.
