If you had to pick the single most dramatic episode in the AI industry in recent years, most people would vote for OpenAI’s boardroom storm. In the span of just five days, the company behind ChatGPT fired its own CEO, cycled through two interim CEOs, nearly had its entire team scooped up by its largest investor, and then brought the man back.
This piece walks through those five days: who Sam Altman is, why his own board fired him, how he came back, and what the storm left behind for OpenAI. If you want the full picture of the company first, you can start with What kind of company is OpenAI.
To set the frame in one line: on the surface it was office politics, but at its core it was the first head-on eruption of the structural tension between OpenAI’s nonprofit mission and its commercial scale.
Who Is Sam Altman
Sam Altman is a co-founder and the CEO of OpenAI. He came out of the startup world, served as president of the well-known startup accelerator Y Combinator, and co-founded OpenAI with a group of others in 2015. Notably, he holds no personal equity in OpenAI, which is rare among tech CEOs and was a deliberate design under the original nonprofit structure.
After ChatGPT set off a global frenzy in late 2022, Altman became almost the face of this wave of generative AI. That is exactly why news of his firing by his own board in late 2023 sent shockwaves through the entire industry overnight.
The Five Days That Rewrote OpenAI
The events mainly unfolded around a weekend in mid-November 2023. The times below are in U.S. West Coast time; some announcements landed the following day in Taiwan time.
| Time | What happened |
|---|---|
| 11/17 | The board abruptly announced Altman’s firing, citing that he had not been consistently candid in his communications with the board. CTO Mira Murati stepped in as interim CEO; chairman Greg Brockman was asked to step down as chairman, and later that day he resigned outright |
| 11/18 | Investors pushed for Altman’s reinstatement; an internal company memo said the firing was not in response to any wrongdoing, and that the issue was a breakdown in communication |
| 11/19 | Talks over Altman’s return briefly collapsed; the board turned to Twitch co-founder Emmett Shear as the second interim CEO |
| Late 11/19–20 | Microsoft CEO Satya Nadella announced that Altman, Brockman, and colleagues would join Microsoft to lead a new AI research team |
| 11/20 | The employee letter went public; ultimately more than 700 employees, about 95% of the entire company, signed it, demanding the board resign and reinstate Altman or they would walk out en masse to join Microsoft |
| Late 11/21 | OpenAI announced an agreement in principle: Altman would return as CEO, with a new initial board made up of Bret Taylor (chair), Larry Summers, and Adam D’Angelo |
| 11/29 | Altman formally returned as CEO and Brockman returned as president; Microsoft gained a non-voting observer seat on the board |
From being fired to reaching an agreement in principle on his return, only about five days passed, and within those five days OpenAI went through three CEOs in total.
Who Fired Him, and Why
The four directors who voted to fire Altman were: co-founder and chief scientist Ilya Sutskever, Quora CEO Adam D’Angelo, and independent directors Helen Toner and Tasha McCauley. Altman and Brockman did not take part in the vote.
The only reason the board gave publicly was that Altman had “not been consistently candid in his communications with the board, hindering its ability to exercise its responsibilities.” It was a heavy statement, but also a vague one: the board never disclosed exactly which communications had gone wrong. That deliberate vagueness is precisely why so many competing accounts of this storm emerged later.
One key design point is worth flagging here: most of OpenAI’s board members held no equity in the company, and their legal duty was to the nonprofit mission rather than to investors or the share price. The design was meant to keep governance from being hijacked by commercial interests, but it also handed a safety- and mission-first board the power to fire the CEO outright.
Ilya Sutskever’s Reversal
The most dramatic personal arc in this storm belongs to Ilya Sutskever. He is a co-founder and the chief scientist of OpenAI, and he was one of the key figures who pushed the concerns into the open and voted to fire Altman.
But three days later, he publicly said on social media that he deeply regretted his participation in the board’s actions, and he signed the employee letter demanding Altman’s reinstatement. After the dust settled, he lost his board seat but stayed on as chief scientist for a while, finally leaving OpenAI in May 2024 and going on to found Safe Superintelligence, a new company focused on AI safety. His departure is often seen as a footnote to the split between OpenAI’s safety camp and its commercial camp.
The Employee Revolt and Microsoft’s Backstop
What truly flipped the situation was the employees.
After the letter went public, the number of signatures kept climbing, and ultimately more than 700 employees, about 95% of the entire company, signed it, demanding that the board resign, that independent directors be reappointed, and that Altman and Brockman be reinstated, or they would resign en masse. Intriguingly, even Mira Murati, who had briefly been appointed interim CEO by the board, and Ilya Sutskever, who had voted to fire Altman, were on the list of signatories.
At the same time, Microsoft played its decisive card. As OpenAI’s largest investor and cloud partner, Microsoft CEO Satya Nadella both brokered behind the scenes and publicly stated he was willing to hire Altman and the entire team to continue their work in a new AI research unit at Microsoft. Employees also said Microsoft had assured them there was a spot for everyone in the new unit.
That effectively backed the board into a corner: if it held firm, OpenAI’s talent could very well migrate wholesale to its own largest investor, leaving the company an empty shell. At that point, the board’s position was untenable.
The Finale: He Came Back, and the Board Was Overhauled
Late on November 21, OpenAI announced that the two sides had reached an agreement in principle and that Altman would return as CEO. The new initial board consisted of three people: former Salesforce co-CEO Bret Taylor as chair, former U.S. Treasury Secretary Larry Summers, and Adam D’Angelo, the only member retained from the old board. Helen Toner, Tasha McCauley, and Ilya Sutskever, who had voted to fire Altman, all left the board.
On November 29, Altman formally returned as CEO and Brockman returned as president. Microsoft obtained a non-voting observer seat on the board, allowing it to sit in on meetings and stay informed but not to participate in votes.
The Aftermath: Independent Review and Governance Reform
After the storm, the new board commissioned the law firm WilmerHale to conduct an independent review, which released a summary in March 2024. The reviewers examined more than 30,000 documents and conducted dozens of interviews, and their conclusions were broadly:
- The core of the firing was a breakdown of trust between the prior board and Altman.
- The review explicitly ruled out several causes that outsiders had speculated about: the firing was not because of product safety, development speed, or company finances, nor because of any misrepresentation to investors, customers, or partners.
- The prior board had “broad discretion” to make the decision, but Altman’s conduct “did not rise to the level that required his removal.”
- The board acted on an overly hasty timeline, without giving advance notice to key stakeholders, without giving Altman a chance to respond, and without anticipating that the move would throw the company into turmoil.
After the review, Altman rejoined the board, and OpenAI added several new directors, introduced new governance guidelines, strengthened its conflict-of-interest rules, and set up a whistleblower hotline.
That storm also planted the seeds for what came later. In 2025, OpenAI completed its corporate restructuring, locking in a two-tier structure in which the nonprofit OpenAI Foundation controls the for-profit OpenAI public benefit corporation (PBC). Many people see the 2023 crisis as one of the distant causes of this restructuring: it demonstrated, in stark terms, just how much control the nonprofit board held over the for-profit entity, and that is exactly what the 2025 restructuring set out to rearrange. It’s worth flagging that this causal relationship is an outside interpretation, not OpenAI’s official verdict on the storm. To see how OpenAI’s valuation and ownership structure have evolved, you can refer to OpenAI Pre-IPO Deep Dive.
What Has Never Been Made Public
For all that has been said about this storm, some key points still have no official verdict, and I’ll honestly flag them here:
- What “not candid” actually refers to: The board never disclosed specific evidence, and WilmerHale only released a summary, not the full report. Former director Helen Toner gave her side’s version in a 2024 interview, for instance saying the board learned of ChatGPT’s imminent launch from Twitter; but that is, after all, the account of a single director, not an official finding.
- The “Q*” rumor: Reuters once reported that researchers had written to the board before the firing, warning of a research breakthrough code-named Q*. That report was single-source, other outlets offered counter-evidence, and OpenAI never confirmed its accuracy; the later independent review further explicitly ruled out safety as the main cause of the firing. So at most it counts as a rumor from the time, and cannot be written as fact.
- Whether “safety vs. commercialization” was the real reason: This is the interpretive frame most media adopted, and the underlying tension did exist; but since the official review ruled out safety and development speed, the more precise way to put it is that this tension was the backdrop to the events, and the official account did not designate it as the direct reason.
Penchan’s Take
Pull these five days back into perspective, and what was really being tested was a governance design.
Back then, OpenAI placed a nonprofit board, one that held no equity, was accountable to the mission, and held a unilateral power to fire, on top of a company that was commercializing at full speed and building the most widely used AI product in the world. When those two forces collided, they set off a storm no one was prepared for. From this angle, the 2025 restructuring that rearranged control was almost an inevitable step after the crisis. For anyone trying to understand this company, these five days were more than tabloid fodder; they offered a rare glimpse into OpenAI’s governance DNA.
Further reading: What kind of company is OpenAI, What kind of company is Anthropic; the safety camp that left OpenAI back then would later become its most direct rival.