Archive · 9 Oct 2026
Fired OpenAI Researchers Challenge Dismissals, Warn of Chilling Effect
OpenAI fired three researchers who worked on AI safety.
The researchers say the company's reasons for firing them do not add up.
They worry that other workers may become afraid to talk about safety or work with outside experts.
OpenAI says the firings were about rules for handling sensitive company information, not about speaking up.
The company says an investigation found a pattern of misconduct, but it has not publicly explained every alleged violation.
Balesni said he was criticized for talking with outside safety groups.
Wang said she had permission to use an executive's email for recruitment and had asked IT to remove that access.
The researchers also say it is important to monitor how advanced AI systems work.
OpenAI said it agreed with their safety recommendations.
Former OpenAI researchers Jasmine Wang, Tomek Korbak and Mikita Balesni disputed the company's reasons for firing them last week.
In an open letter published Thursday, they said the abrupt dismissals could make colleagues afraid to raise safety concerns or work with outside experts.
OpenAI says the firings were for alleged violations of policies on sensitive information and that an investigation found a “pattern of misconduct”; it has not detailed all alleged violations.
The researchers deny mishandling sensitive information; Balesni said he was criticized for speaking with outside safety organizations, while Wang said her access to an executive's email was authorized for recruitment.
OpenAI rejected the claim that the researchers were fired for speaking out, and said in an internal memo that it agreed with their safety recommendations.
- Who
- Former OpenAI researchers Jasmine Wang, Tomek Korbak and Mikita Balesni, and OpenAI.
- What
- The researchers challenged their firings and warned the dismissals could discourage open discussion of AI safety; OpenAI says they were fired for alleged sensitive-information policy violations.
- When
- The researchers were fired last week; their open letter was published Thursday. No calendar date is specified.
- Where
- At OpenAI; no physical location for the dispute is specified.
- Why
- The researchers dispute the company's explanation and say the dismissals could chill safety discussions. OpenAI says the decisions followed an investigation into alleged policy violations.
Researchers' account
OpenAI's account
Reasons for the dismissals
Researchers' account
The researchers dispute the allegations and say they acted in good faith. Balesni said his external safety-group communications were coordinated with company leadership; Wang said her email access was authorized for recruitment.
OpenAI's account
OpenAI says the researchers were fired for violating policies on sensitive company information, following an investigation that found a pattern of misconduct.
Whether safety advocacy prompted the firings
Researchers' account
The researchers suggest their safety work or external communications contributed to the dismissals and warn the firings could deter colleagues from speaking openly.
OpenAI's account
OpenAI says the researchers were not fired for raising safety concerns and stated that it does not terminate employees for doing so.
Impact on company culture
Researchers' account
The researchers say abrupt dismissals and communications about them have made former colleagues afraid to speak and could weaken collaboration with outside experts.
OpenAI's account
OpenAI rejected the claim that the firings were retaliation for safety concerns and said it agreed with the researchers' recommendations.
- Researchers
- Jasmine Wang, Tomek Korbak and Mikita Balesni
- Employer
- OpenAI
- Timing
- The researchers were fired last week; their open letter was published Thursday.
- OpenAI's stated basis
- Alleged violations of policies governing sensitive company information.
- Investigation
- OpenAI said an investigation found a “pattern of misconduct” but did not publicly detail all alleged violations.
- Balesni's account
- He said he was told OpenAI no longer trusted him because he spoke too much with third-party safety organizations; he denied sharing company intellectual property.
- Wang's account
- She said her access to an executive's email was delegated for recruitment and that she asked IT to revoke it when it was no longer needed.
- Safety concern
- The researchers said monitoring advanced AI systems, including their chain-of-thought, is important.
