The OpenAI Board Crisis, November 17–22, 2023

---

The Setup: A Strange Governance Structure

OpenAI was founded in 2015 as a nonprofit research lab. By 2019, it was clear that training frontier AI models required more computing power — and more money — than any donor was willing to give. So OpenAI created a for-profit subsidiary, controlled by the nonprofit parent. Investors in the for-profit subsidiary could earn returns, but only up to a cap. After the cap, all further profits would flow to the nonprofit. The nonprofit board, not the investors, was in charge.

This structure meant that a small board of directors — six people at the time of the crisis — controlled one of the most valuable companies in the world:

• Sam Altman, the CEO
• Greg Brockman, the president and board chair
• Ilya Sutskever, the chief scientist and a co-founder
• Adam D’Angelo, the CEO of Quora
• Helen Toner, a researcher at Georgetown University’s Center for Security and Emerging Technology
• Tasha McCauley, a tech entrepreneur and adjunct senior management scientist at RAND

The structure was designed so that the board could fire the CEO if it believed he was putting commercial interests ahead of the nonprofit’s mission. The structure did not require the board to consult investors, employees, or partners before doing so. This design choice, which seemed like a feature when it was created, became a serious problem in November 2023.

The board was uneasy. The details of what exactly made them uneasy are contested, but the broad shape of the concern was this: the board believed that Altman was not always honest with them, and they worried about the pace at which OpenAI was commercialising technology that the board believed could be dangerous.

---

Friday, November 17: The Firing

On Thursday, November 16, Sam Altman spoke at the APEC CEO Summit in San Francisco. That evening, he received a text message from the OpenAI board asking him to join a Google Meet video call the next day.

On Friday, November 17, Altman was in Las Vegas for the Formula 1 Grand Prix weekend. Around noon Pacific time, he joined the Google Meet from Las Vegas. The entire board was on the call except Greg Brockman. Ilya Sutskever, the chief scientist, told Altman that he was being fired, and that the news would be made public imminently. Altman was given roughly five to ten minutes of warning before the announcement went out.

OpenAI published a blog post. The key sentence was: “Mr. Altman’s departure follows a deliberative review process by the board, which concluded that he was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities. The board no longer has confidence in his ability to continue leading OpenAI.”

That sentence — “not consistently candid in his communications with the board” — would be analysed, mocked, and agonised over for the next five days. Greg Brockman was removed as board chair, told he would also be removed from the board, and resigned in protest. Three senior researchers — Jakub Pachocki, Aleksander Madry, and Szymon Sidor — also departed. The board named Mira Murati, OpenAI’s Chief Technology Officer, as interim CEO.

Microsoft was not meaningfully consulted. Satya Nadella, the CEO of Microsoft — whose company had $13 billion invested in OpenAI and whose entire consumer AI strategy was built on OpenAI’s technology — was given roughly one minute of advance notice before the public announcement. Microsoft’s Copilot products, integrated into Office, Windows, and Bing, depended on OpenAI’s models, which made the firing a direct threat to Microsoft’s strategy. Within hours, OpenAI’s investors — Thrive Capital, Sequoia Capital, Microsoft — were furious and began pressuring the board to reverse the decision.

---

Saturday, November 18: The Negotiations

On Saturday, the board and Altman began negotiations for his return. According to reporting by Bloomberg and The New York Times, the board “agreed in principle” to resign and reinstate Altman and Brockman. But the agreement fell apart. The board missed a deadline. Members waffled. The negotiations collapsed by the end of the day.

By Saturday evening, the most likely outcome seemed to be that Altman would return, the board would be reconstituted, and the crisis would end. That is not what happened.

---

Sunday, November 19: Emmett Shear and the Microsoft Move

On Sunday morning, the situation changed. The OpenAI board named Emmett Shear — the co-founder and former CEO of Twitch (Amazon’s video-game streaming platform) — as the new interim CEO, replacing Mira Murati, who had held the role for less than 48 hours. Shear reportedly described his mandate as “to stabilize and reform within 30 days.”

Then Satya Nadella made his move. On Sunday evening, Nadella announced that Altman, Brockman, and their colleagues would join Microsoft to lead a new advanced AI research subsidiary. The implication was clear: Microsoft was effectively hiring OpenAI’s leadership, and the OpenAI board had lost.

---

Monday, November 20: The Employee Revolt

On Monday, the OpenAI board faced the most damaging consequence of its decision. More than 700 of OpenAI’s roughly 770 employees — about 95% of the company — signed an open letter demanding that the board resign and reinstate Altman. The letter threatened that the employees would resign en masse and join Microsoft’s new AI unit.

The signatories included Ilya Sutskever. This was a striking reversal. Sutskever had been one of the board members who voted to fire Altman on Friday. By Monday, he was signing a letter demanding the board’s resignation. He tweeted: “I deeply regret my participation in the board’s actions. I never intended to harm OpenAI. I love everything we’ve built together and I will do everything I can to reunite the company.”

The employee revolt was the decisive moment. The board could perhaps have survived the investor pressure. It could perhaps have survived Microsoft’s move. It could not have survived the loss of 95% of its employees.

---

Tuesday, November 21: The Deal

On Tuesday evening, the deal came together. OpenAI confirmed that it had reached an “agreement in principle” for Altman to return as CEO. The new “initial board” would consist of three people:

• Bret Taylor — the former co-CEO of Salesforce and former chair of Twitter’s board, as chair
• Larry Summers — the former US Treasury Secretary under President Bill Clinton, as a director
• Adam D’Angelo — the CEO of Quora, the only holdover from the old board

The board would expand over time to as many as nine directors. Microsoft would receive a non-voting observer seat — informed of major decisions but unable to vote on them — a direct response to its having been blindsided.

---

Wednesday, November 22: Q* and the Aftermath

On Wednesday, November 22, OpenAI published a formal announcement. Altman was back as CEO. The new board was in place. The crisis was over.

But the story was not quite finished. On the same day, Reuters published an exclusive report that the firing had been “precipitated by a letter” from OpenAI researchers to the board warning of an AI breakthrough called “Q”* (pronounced “Q-Star”). The letter described Q* as a system that could solve certain math problems that current AI systems could not. Some researchers believed Q* might be a step toward artificial general intelligence. OpenAI never confirmed the Reuters report. The Q* story became one of the enduring mysteries of the crisis.

---

What the Board Said Later

For six months, the board members who had fired Altman said very little. Helen Toner broke her silence in May 2024, on The TED AI Podcast. Her account was the most detailed public explanation from the board’s side.

Toner said that the firing was not primarily about AI safety, as much of the press had assumed. It was about trust. She said that Altman had “outright lied” to the board on multiple occasions — about safety concerns, about board composition, and about his attempts to push her off the board. She said that two senior executives at OpenAI had reported to the board that Altman had been dishonest with them as well. Sam Altman and OpenAI did not give a detailed rebuttal to Toner’s specific allegations.

---

Ilya Sutskever’s Trajectory

Of all the figures in the crisis, Ilya Sutskever had the most interesting trajectory. He had been one of the board members who voted to fire Altman on Friday. By Monday, he had signed the open letter demanding the board’s resignation. On Tuesday, when Altman returned, Sutskever was reportedly emotional.

For six months after the crisis, Sutskever stayed at OpenAI but was rarely seen in public. On May 14, 2024, he announced that he was leaving OpenAI. In June 2024, he announced that he had co-founded Safe Superintelligence Inc. (SSI) with Daniel Gross and Daniel Levy. SSI’s stated mission was to build safe superintelligence — to pursue the goal of artificial general intelligence, but with safety as the primary focus. In September 2024, SSI raised $1 billion at a valuation of approximately $5 billion — an extraordinary valuation for a company with no products, no revenue, and no timeline for delivering either.

---

What Changed: The Governance Aftermath

In January 2024, Microsoft’s Dee Templeton joined the OpenAI board as a non-voting observer. (Microsoft and Apple both gave up their observer seats in July 2024, citing antitrust scrutiny, but Microsoft’s influence over OpenAI remained substantial.) In March 2024, OpenAI expanded its board — Sam Altman was reinstated as a director, and Sue Decker (former Yahoo CFO), Nicole Seligman (former Sony executive), and Fidji Simo (CEO of Instacart) joined. In October 2024, OpenAI raised $6.6 billion at a post-money valuation of $157 billion, led by Thrive Capital with Microsoft and Nvidia participating.

The most significant governance change came in 2025. OpenAI announced its intention to convert its for-profit arm into a Delaware Public Benefit Corporation (PBC) and to remove the profit caps that had been a defining feature of the original structure. The announcement drew opposition from Elon Musk (who filed a lawsuit alleging that OpenAI had abandoned its nonprofit mission), from public charities, from former employees, and from the California Attorney General, who opened an investigation. In May 2025, OpenAI reversed course on one aspect: the nonprofit parent would retain control of the for-profit PBC. In October 2025, OpenAI completed the conversion, with the nonprofit receiving an equity stake worth approximately $130 billion.

The conversion was a direct downstream consequence of the November 2023 crisis. The original nonprofit-controlled structure had been designed to ensure that the nonprofit’s mission would prevail in any conflict with commercial interests. The crisis revealed that this structure was unworkable: when the board tried to exercise its authority, the result was chaos, employee revolt, and the effective removal of the board.

---

What It All Meant

The OpenAI board crisis was the most dramatic governance event in the history of the AI industry. In five days, the most important AI company in the world fired its CEO, lost 95% of its employees to a threatened mass resignation, hired a new CEO, lost him, hired the old CEO back, and replaced its board. The crisis exposed the contradictions at the heart of OpenAI’s structure — a nonprofit mission controlled by a small board, funded by billions of dollars of profit-seeking investment, partnered with the world’s largest software company — and revealed that those contradictions could not be papered over indefinitely.

The crisis also exposed the central tension in the AI industry: the tension between building AI quickly and building it safely. The original board, especially Toner, McCauley, and Sutskever, represented the safety side. Altman represented the speed side. The crisis did not resolve the tension. It simply moved it from inside OpenAI to the broader industry, where it continues to play out.

The lesson of the crisis is not that the board was right or that Altman was right. It is that the structure for governing powerful AI systems — inside companies, and eventually inside governments — is still being worked out. The OpenAI board crisis was one early attempt to deal with this problem. It failed. Future attempts will need to do better.

---

---

What would change if more people understood the story behind the OpenAI board crisis? Who benefits from the current state of affairs, and who is left out? The conversation is worth having — with colleagues, with students, with anyone who uses technology without thinking about where it comes from.