OpenAI Denies Safety Researcher Dismissals Were Retaliation
OpenAI says it dismissed three safety researchers for violating sensitive-information policies, not for raising safety concerns, and will announce a third-party evaluation deal soon.
OpenAI issued an official statement on X on October 9, 2026, saying its recent dismissal of three safety researchers was not retaliation for raising safety concerns. The company said its investigation found the dismissed researchers violated clear policies on handling sensitive information.
According to aitimes.com, the statement also said OpenAI has never fired employees for raising safety concerns and will maintain that practice.
OpenAI added that it agrees with the argument, made in a letter from the dismissed researchers, that industry-wide efforts are needed to maintain the monitoring capability of frontier models. The company said it is finalizing a contract with a third-party safety evaluation organization and will announce details within weeks.
Background
Last week, OpenAI parted ways with researchers Jasmine Wang, Tomek Korbak, and Mikita Balesni. On October 8, the three sent an open letter to OpenAI's board and safety committee denying the alleged leaks. Earlier, on September 23, CEO Sam Altman urged the international community to establish AI safety standards in a speech to the UN Security Council, and said OpenAI would postpone its planned 2026 IPO over AI safety concerns.
Quick answers
Why did OpenAI dismiss the three safety researchers?
OpenAI said its investigation found they violated clear policies on handling sensitive information, and that the dismissals were not retaliation for raising safety concerns.
What did the dismissed researchers say?
On October 8, Jasmine Wang, Tomek Korbak, and Mikita Balesni sent an open letter to OpenAI's board and safety committee denying the alleged leaks.
What is OpenAI's next step on safety evaluation?
OpenAI said it is finalizing a contract with a third-party safety evaluation organization and will announce details within weeks.