Three fired OpenAI safety researchers say the reasons don't add up

The news

Jasmine Wang, Tomek Korbak and Mikita Balesni, three safety researchers OpenAI fired at the start of October, published an open letter denying the company's claim that they mishandled sensitive information outside established procedures, and warned of a chilling effect on the company's culture, TechCrunch reported on October 8.

Why it matters

How a frontier lab treats internal safety critics is a test of whether its safety commitments survive pressure. Regulators in Washington and Brussels are now asking the same labs about incidents.

The details

OpenAI told NPR the three violated clear policies on handling sensitive information more than once, and that the mishandling went beyond their work with an outside evaluation group. An internal memo from a research leader said the company strongly agreed with the researchers' recommendations and does not terminate employees for raising concerns.

Wang wrote on X that she was told she was fired for accessing an executive's email. She said she had been given delegated access to that inbox for recruiting, asked IT to remove it, and reported opening a sensitive email by mistake within minutes.

Separately, David Robinson, who wrote safety reports for OpenAI's major launches before resigning, told NPR that iterative deployment worked for chatbots but not for agents that act in the world.

What we could not check

• We could not verify either side's account of the internal investigation.
• OpenAI has not published the policies it says were violated.