Fired OpenAI Safety Researchers Deny Misconduct and Warn of Chilling Effect

Jasmine Wang, Tomek Korbak and Mikita Balesni sent the letter Thursday to three OpenAI bodies: the Safety and Security Committee, the Safety Advisory Group and the Mission Advisory Council. In it, they deny wrongdoing and push back on the characterization of their exit.

OpenAI dismissed the three last week, saying they breached company policy through the "accessing and handling sensitive company information" and alleging they shared confidential material with a third-party AI safety organization. The researchers say they did not handle sensitive information outside established company procedures.

The letter argues that the dismissals and the communications surrounding them have made remaining staff wary. The researchers wrote that colleagues are now afraid to speak and operate in ways that were, until last week, a routine part of working at the company.

"AI is not a normal technology, and OpenAI is not a normal company," they wrote, adding that employees working on safety tend to identify risks earlier than others and depend on close collaboration with outside experts to address them.

That freedom to work with external specialists, and the clearly defined internal processes that support it, is itself a core safety mechanism, the researchers argued. Their letter states that the ability to operate without fear is essential to the function.

The dispute places a public spotlight on how OpenAI governs its safety work and how it handles staff who engage with outside researchers, with the three former employees warning of ripple effects across the company's internal culture.