OpenAI has let go of three people who worked on its safety and alignment teams. The Wall Street Journal says the staff – identified as Jasmine Wang, Tomek Korbak and Mikita Balesni – passed confidential material to an outside AI‑safety group. OpenAI confirmed that an internal investigation found violations of its rules on handling sensitive information, but did not name the employees. The company has not said what data was shared or which organization received it.
Who left and why it matters
Korbak served as OpenAI’s technical contact for two external groups, METR and Redwood Research, which examine how OpenAI’s AI agents bypass security controls. Wang and Balesni worked on alignment, the effort to keep AI behavior in line with human values. An anonymous X account later reported that a fourth safety researcher, David Robinson, left the firm shortly after the firings. All four had spoken publicly about AI risks in September, with statements ranging from concerns about a rapid self‑improving AI race to a claim that the chance of AI wiping out humanity is above ten percent.
Why it matters
The firings show that OpenAI is tightening control over how internal safety data is shared, which could affect collaborations with outside researchers. For people following AI risk debates, the incident highlights a tension between transparency on safety concerns and protecting proprietary information. It remains unclear what specific data was leaked or how the departures will impact OpenAI’s ongoing safety projects.
