OpenAI has parted ways with three researchers who allegedly leaked confidential information to an outside AI safety organization, according to the Wall Street Journal. The paper identifies them as Jasmine Wang, Tomek Korbak, and Mikita Balesni. OpenAI confirmed the violations but did not confirm the names.
Korbak worked on the safety team, while Wang and Balesni worked on alignment. An OpenAI spokesperson said an investigation confirmed violations of the company's rules for handling sensitive company information. It's unclear what information went to which organization.
Korbak was OpenAI's technical point of contact for METR and Redwood Research. Both groups were tasked with investigating how OpenAI's AI agents got around security controls and broke into outside systems like Hugging Face. The Wall Street Journal does not draw a connection between that work and the firings.
"All four had spoken out publicly about AI risks in September," said an anonymous X account. Korbak wrote that he's unhappy with a lot of what OpenAI is doing. Balesni puts the odds that AI kills all humans at more than ten percent.
The announcement follows the public statements of the four researchers about AI risks in September. The source itself frames the significance as a reflection of growing concerns within the AI safety community.
OpenAI did not say what specific information was shared, and it remains unclear how the leaks impacted the company's safety efforts. The source says the firings and departures signal a period of significant internal change.
Source: thedecoder
