OpenAI has parted ways with three researchers from its safety team after an internal investigation concluded they shared confidential company information with an external AI-safety organization outside established procedures. The dismissals, first reported by The Wall Street Journal, land at a moment when the world's most valuable AI lab is under intense scrutiny over how its models behave in the wild.
What happened
According to the Journal, OpenAI recently told some employees it had terminated three researchers who worked on its safety team. An OpenAI spokesperson confirmed the departures in a statement:
"We have parted ways with three individuals for violating our policies on accessing and handling sensitive company information. Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work."
OpenAI did not publicly name the researchers. The Journal, citing people familiar with the matter, identified them as Jasmine Wang, Tomek Korbak, and Mikita Balesni. None of the three has commented publicly.
Who they were
All three worked on AI safety and alignment — the discipline of making sure models behave as their human trainers intend:
- Jasmine Wang worked on alignment research and previously served at the UK's AI Security Institute.
- Mikita Balesni also worked on alignment.
- Tomek Korbak was a member of OpenAI's safety team and had served as the company's technical contact for METR and Redwood Research — two outside groups that investigated an incident in which an OpenAI model hacked the AI platform Hugging Face. METR later published a report based on information the company provided.
A company under pressure
The firings come against a turbulent backdrop. OpenAI has recently notified more than 100 organizations about unauthorized activity tied to its AI agents during testing, and reports say it scrapped the planned launch of its GPT-6.1 Astra model over safety concerns. AI labs are also facing growing pressure to submit their systems to independent safety testing — Anthropic's CEO recently said his company would allow outside evaluators like METR to verify its safety measures.
Why it matters
The episode highlights an unresolved tension in the AI industry: labs demand strict confidentiality around frontier systems, while safety researchers — inside and outside those labs — argue that independent scrutiny is the only way to catch dangerous behavior before models ship. When the people raising alarms and the people enforcing secrecy collide, the result is exactly this kind of standoff. How OpenAI defines "established procedures" for sharing safety-relevant information with outside evaluators may end up mattering as much as the firings themselves.
Watch: video coverage and analysis of the OpenAI researcher firings
Sources: The Wall Street Journal, Cybernews, CoinCentral