OpenAI Fires Three Safety Researchers Amid Model-Design Leak Dispute
OpenAI said an internal investigation found that Tomek Korbak, Mikita Balesni and Jasmine Wang violated sensitive-information policies, and the company dismissed all three. The Wall Street Journal, as summarized by Techmeme and The Next Web, reported that investigators suspected the disclosure of internal model designs to Ryan Greenblatt through disappearing Signal messages, while the researchers denied sharing confidential information. The dispute followed reporting about Astra, a model described as more capable but exposing less of its chain of thought, and the researchers said they had advocated closer cooperation with external AI-safety groups.
The case brings OpenAI’s internal information controls into conflict with external scrutiny of model designs and chain-of-thought transparency.

“I was told that there was a breach of trust because I interacted with a third party through unsanctioned channels,” Tomek Korbak said.
Korbak is one of three OpenAI safety researchers fired last month. He spoke to The Wall Street Journal for a story by Maxwell Zeff, Keach Hagey and Berber Jin, published on Thursday.
OpenAI’s security team asked the three to open up their personal phones during a leak investigation, the Journal reported. Investigators named AI safety researcher Ryan Greenblatt as the suspected leaker. He denies it. TNW has not independently verified the account.
#OpenAI#Tomek Korbak#Mikita Balesni#Jasmine Wang#Ryan Greenblatt#Astra#Signal#METR#Redwood Research#Hugging Face