Ilya Sutskever Stands Up For His Role In Sam Altman’s OpenAI Firing: ‘I Didn’t Want It To Be Destroyed’

Elon Musk The case against OpenAI and Microsoft entered its final stage on Monday, with testimony from Microsoft CEO Satya Nadella, former OpenAI chief scientist Ilya Sutskever, and current OpenAI chairman Bret Taylor.

Sutskever drew attention, revealing an ownership stake in OpenAI’s $850 billion profit arm that is currently worth about $7 billion. That makes him one of the largest known shareholders of OpenAI. At the beginning of the trial, OpenAI president Greg Brockman admitted for the first time that he owns OpenAI shares worth 30 billion dollars.

Brockman was one of the original founders of the research lab, and Sutskever joined soon after, turning down an offer of $6 million in annual compensation from Google. Brockman said he and Sutskever were “joined at the waist,” until Sutskever helped lead the ouster of Sam Altman as CEO of OpenAI in 2023. Sutskever had helped gather evidence to show Altman’s history of alleged fraud, and even helped write a memo to the board. Although they tried to mend the relationship, Sutskever has since parted ways with Brockman and Altman, an OpenAI attorney said Monday.

Sutskever, who appeared in court wearing a shirt and pants, the first male witness to testify without a suit jacket, appeared disheartened by his lack of involvement with OpenAI. (He left and created a competitive AI lab in 2024.) “I felt a lot of ownership of OpenAI,” he said at one point on Monday. “I felt like I was putting my life into it, and I just cared about it, I didn’t want it to be destroyed.

Sutskever’s testimony reinforced Musk’s argument that Altman is not the right person to lead an AI lab that could create general artificial intelligence. In addition, Sutskever talked about how the superalignment team he helped lead, focused on the security of future models, is doing the most important work for OpenAI “for a long time.” The group disbanded in May 2024, shortly after Sutskever left the company.

But Sutskever also added to OpenAI’s defense that Musk never negotiated any special promises when sponsoring the OpenAI nonprofit. Musk’s allegations that such obligations existed and that Altman and Brockman breached them in pursuit of a profitable for-profit arm are at the heart of his claims in the lawsuit. Sutskever said OpenAI needs “a lot of dollars” to build a computer as big as the human brain, and while soliciting donations has had “reasonable success,” becoming a for-profit was the consensus way forward.

“I would describe it as the difference between an ant and a cat,” Sutskever said in response to a question from U.S. District Judge Yvonne Gonzalez Rogers about how much computing has helped OpenAI scale. “If there is no money, there is no super computer.”

Finally, Sutskever, a prominent AI scientist who paints in his spare time, testified for about an hour, never making eye contact while on the witness stand.

Musk’s legal team failed to seek to treat Sutskever as a hostile witness because of his financial stake in OpenAI. But Gonzalez Rogers agreed to give lawyers for both Musk and OpenAI more leeway in their questioning of Sutskever because of what he described as his “unique position” in the case.

Blip

Much of Monday’s testimony focused on the well-covered events of Altman’s ouster and reinstatement as CEO in November 2023. Nadella described Sutskever and the other board members who fired Altman as “newbies” and insisted he was “not clear” about the lack of integrity that led to their decision. Nadella also acknowledged during his testimony that he and his colleagues discussed about 14 board members who would join OpenAI if Altman returned, including at least two the Microsoft team voted for and one who joined later. Nadella described Microsoft’s input as suggestions.

Sutskever said he supports firing Altman because “an environment where management doesn’t have the right information” is “not conducive to achieving any major objective.” But he criticized his colleagues on the board for rushing the process, being inexperienced, and accepting “legal advice that was not good at all.”

Microsoft’s bet

In his lawsuit, Musk accused Microsoft of helping turn OpenAI into a more money-making machine than Musk intended. Nadella testified that Microsoft initially supported OpenAI with discounted cloud computing but could no longer do so “once the bill started to mount.” The profit arm that Microsoft invested in, to get a financial return, was very attractive.

But as the years went on and the bills kept mounting, Microsoft wanted more from the partnership. Microsoft “will lose 4 billion next year!!!” Nadella raved in a 2022 email to his bosses about the OpenAI collaboration. He called for a new deal that guaranteed Microsoft would get AI “know-how” from the start, which he kept spelling as “Turn on AI.”