The OpenAI Coup — Altman Fired, Reinstated in Five Days

On the afternoon of Friday, 17 November 2023, OpenAI published a short post announcing that "Sam Altman will depart as CEO and leave the board of directors." Five days later he was on his way back.

17 November — The Motion

Since its 2015 founding, OpenAI had operated under a peculiar governance structure: the non-profit OpenAI, Inc.—a 501(c)(3)—was, in its own words, "the overall governing body for all OpenAI activities", sitting above a capped-profit operating company, OpenAI Global, LLC. In autumn 2023 the board comprised six members—Sam Altman, Greg Brockman, Ilya Sutskever (chief scientist), Adam D'Angelo (CEO of Quora), Tasha McCauley, and Helen Toner (Georgetown's Center for Security and Emerging Technology).

The announcement gave one reason, and only one: a board review "concluded that he was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities." Chief technology officer Mira Murati was named interim CEO, effective immediately. Greg Brockman was stepped down as chairman the same day and told he would report to the CEO; within hours he resigned outright.

The substance of the allegations was not disclosed, then or since. What emerged later—safety concerns, disputed accounts of Altman's dealings with individual directors, the pace of commercialisation—was vague and lightly evidenced. On Saturday morning COO Brad Lightcap told staff in a memo obtained by CNBC that "the board's decision was not made in response to malfeasance or anything related to our financial, business, safety, or security/privacy practices," and that "this was a breakdown in communication between Sam and the board." The stated reason had thus been narrowed to almost nothing.

17–22 November — The Revolt

Across the weekend, events moved fast.

Talks to reinstate Altman ran through Saturday and Sunday and collapsed. Late on Sunday 19 November the board replaced Murati with former Twitch CEO Emmett Shear—OpenAI's third chief executive in three days. Within hours Microsoft closed the exit: Satya Nadella wrote that Altman and Brockman, "together with colleagues, will be joining Microsoft to lead a new advanced AI research team." Altman later confirmed the timing in his own words: "when i decided to join msft on sun evening, it was clear that was the best path for me and the team."

On Monday 20 November an open letter to the board began circulating among staff. It demanded that all sitting directors resign and that Altman and Brockman be reinstated, and warned that the signatories would otherwise leave for the new Microsoft unit, which had told them there were positions for every OpenAI employee. The count climbed through the day past 500, then 700, and is recorded at 745 of roughly 770 employees. Among the signatures was Ilya Sutskever's—who had reportedly led the move against Altman.

Late on Tuesday 21 November, US Pacific time, OpenAI announced an agreement in principle for Altman to return as CEO under a reconstituted board. The arrangement was formalised on 29 November: Altman as CEO, Murati back as CTO, Brockman as president, and an initial board of three—Bret Taylor (chair, formerly co-CEO of Salesforce), Larry Summers (former US Treasury Secretary), and Adam D'Angelo, the only holdover. Sutskever, Tasha McCauley, and Helen Toner were off the board; Microsoft took a non-voting observer seat.

Who Actually Governs OpenAI

In the form of a corporate power struggle, the five days were ordinary enough: dismissal, employee revolt, reinstatement. But because the object was the best-capitalised laboratory at the AI frontier, the episode left a wider mark.

• Who actually governs AI development? The named board, or the engineers who run the product and the large investors? The answer was unmistakably the latter.
• The split between safety and acceleration camps. The faction that drove the dismissal (Toner, McCauley, Sutskever) was identified with caution about AI risk. Their complete defeat tilted the balance inside the industry in one direction.
• Microsoft's gravitational pull. Microsoft could, in a matter of days, have absorbed OpenAI's core. OpenAI's "independence" was no longer separable from Microsoft's presence.

Sutskever announced his departure in May 2024 and co-founded Safe Superintelligence Inc. the following month. Over the following years, the major figures of the safety camp left OpenAI one by one.

The structure that made the coup possible did not survive it. Bret Taylor's board spent two years rebuilding it, and on 28 October 2025 OpenAI completed a recapitalisation: the non-profit, renamed the OpenAI Foundation, kept control of a renamed operating company, OpenAI Group PBC, while holding roughly 26 per cent of its equity, then worth some $130 billion; Microsoft's stake was fixed at about 27 per cent. Larry Summers resigned from the board on 19 November 2025, days after the release of his correspondence with Jeffrey Epstein. Taylor and D'Angelo remain.

The five-day affair will be remembered as the moment when the real governance structure of the AI industry was, for once, laid bare in daylight.