OpenAI Fires Three Over Data Leak, Not Safety

Jasmine Wang, Tomek Korbak, and Mikita Balesni lost their jobs at OpenAI last week after the company accused them of sharing confidential information with an outside AI safety organization without authorization. OpenAI framed the dismissals as a straightforward policy violation, saying the three had mishandled sensitive material outside approved channels. The researchers see it differently, and on Thursday they pushed back in an open letter addressed to OpenAI's Safety and Security Committee, its Safety Advisory Group, and its Mission Advisory Council.

Their argument isn't just about clearing their own names. Wang, Korbak, and Balesni contend that how OpenAI has handled their firing, both internally and in public statements, has left remaining staff rattled about what they can and can't say or do. That matters for a company whose safety researchers often need to work closely with outside experts to spot and flag risks before they become real problems. The letter lands at a moment when OpenAI's internal culture around dissent and disclosure is already under scrutiny, and it raises a pointed question about whether the company's own safety staff now feel safe speaking up at all.