OpenAI Fires Researchers for Leaking Information

Inspector Gadget

After months of leaks, OpenAI has apparently fired two researchers who are said to be linked to company secrets going public.

As The Information reports based on insider knowledge, the firm has fired researchers Leopold Aschenbrenner — said to be a close ally of embattled cofounder Ilya Sutskever — and Pavel Izmailov.

Both men had worked on OpenAI's safety team, per the report, though most recently, Aschenbrenner had been working on its so-called "superalignment" efforts while Izmailov was tasked with researching AI reasoning. We've reached out to OpenAI and Microsoft, its major investor, for further context.

While it's hazy what information the men leaked, it's been clear for months now both that OpenAI has become something of a leaky ship and that it was looking to do something about it.

Back in February, the blog MSPowerUser spotted a since-deleted job listing for an in-house detective to help with "analyzing anomalous activities, promoting a secure culture, and interacting with various departments to mitigate risks."

Notably, the "insider risk investigator" listing had, as the Wayback Machine indicates, been up since at least mid-January, which if you recall was only a few months after the Thanksgiving massacre that saw the firing of CEO Sam Altman before his subsequent return.

Rocky Road

That same failed coup, given a vulgar moniker by folks at Microsoft, also saw historic leaks from within the company, including claims that Altman was fired over fears of a purportedly mega-powerful and secretive AI model called Q*.

Because this latest report is based on — yes — a leak, it's not clear how well the leak crackdown is going.

It also remains uncertain what implications the firings will have on the rest of the firm, but Aschenbrenner's purported closeness with Sutskever seems salient. As we learned during the failed coup, the cofounder had been instrumental in Altman's ouster, and has been said to be in a sort of limbo within the company ever since. Following that debacle, fellow cofounder Elon Musk seemed to suggest that he was looking to poach Sutskever, too.

While Altman said following his return as CEO that he has "zero ill will" towards Sutskever, his protegé's firing could be seen as evidence to the contrary — and this reporting, if true, could signal turbulence under the calm facade the company has projected since the coup.

More on OpenAI: OpenAI and Meta Reportedly Preparing New AI Models Capable of Reasoning