Artificial Intelligence
OpenAI dismisses three safety researchers, and a fourth leaves
OpenAI ended ties with three researchers after an investigation into leaks of sensitive information, while a fourth safety researcher was reported to have left soon afterwards.
- The Wall Street Journal said OpenAI parted ways with Jasmine Wang, Tomek Korbak and Mikita Balesni.
- An OpenAI spokesperson said an investigation found breaches of company rules for handling sensitive information, and OpenAI did not confirm the names.
- Korbak had been OpenAI's technical contact for METR and Redwood Research, which were looking at how OpenAI's AI agents got around security controls.
- An anonymous X account said David Robinson left OpenAI shortly afterwards.
- The article said all four had spoken publicly in September about risks from advanced AI.
The Wall Street Journal reported that OpenAI had parted ways with Jasmine Wang, Tomek Korbak and Mikita Balesni. The report said the company had linked the action to leaks of confidential information to an outside AI safety group.
An OpenAI spokesperson said an inquiry had found breaches of the company's rules on handling sensitive material. OpenAI did not confirm the names, and the article said it was not clear what information had gone to which organisation.
Korbak had been the technical point of contact at OpenAI for METR and Redwood Research. Those groups were testing how OpenAI's AI agents could get around security controls and enter outside systems, and the Wall Street Journal did not say that this work was connected to the firings.
An anonymous X account said a fourth safety researcher, David Robinson, left OpenAI soon after. The article said all four had spoken publicly in September about AI risk, including claims that development was moving too fast and that advanced systems could pose serious harm.
Named in this story
People
- Jasmine Wang
- was one of the three researchers named in the report
- Tomek Korbak
- was named as one of the three researchers and worked on the safety team
- Mikita Balesni
- was named as one of the three researchers and worked on alignment
- David Robinson
- was said by an anonymous X account to have left shortly afterwards
Companies
- OpenAI
- parted ways with three researchers after an internal investigation
Organisations
- Wall Street Journal
- identified the researchers in its reporting
- METR
- had Korbak as OpenAI's technical point of contact
- Redwood Research
- worked with OpenAI on testing how its agents bypassed controls
How the source tells it
Neutral and reportorial, with the main claims attributed to the Wall Street Journal, an OpenAI spokesperson and an anonymous X account.
No emotive framing was found in the original.