AI news · OpenAI · safety
OpenAI says fired safety researchers broke information rules
The three researchers published a letter saying their dismissals chill safety work; OpenAI answered that they mishandled sensitive information.
OpenAI said on Friday 9 October 2026 that it fired safety researchers Mikita Balesni, Tomek Korbak and Jasmine Wang after an investigation found they broke “clear policies on handling sensitive information”, CBS News reports. The three were let go the week before, as this section reported on 2 October.
OpenAI’s post answered a letter the three made public on Thursday. They asked the company to keep working with outside safety groups and to protect the ability to monitor model behaviour, and warned that the firings leave former colleagues afraid to speak, NPR reports. Balesni wrote on X that he was told OpenAI no longer trusted him over his contacts with outside safety organisations, and said he never shared company IP. Korbak was the technical contact for the METR and Redwood Research investigation into the Hugging Face breach.
OpenAI says its investigation found a breach of trust beyond what the letter describes, without giving details, The Verge reports. It also says it agrees with the letter’s recommendations and still plans to bring in outside evaluators, CBS News reports.
For buyers of frontier models, the dispute puts one question on the checklist: how much access outside evaluators get to a lab’s incident records.
Earlier on this story: OpenAI fires three safety researchers over shared information
Sources
- CBS News: OpenAI defends firing safety researchers. Checked
- NPR: Fired OpenAI employees allege retaliation over safety concerns. Checked
- The Verge: OpenAI doubles down on decision to fire three AI safety researchers. Checked
Alexandre S.
Alexandre S. is a creative ad strategist and copywriter with ten years' experience in digital marketing across multiple channels.