(San Francisco) Around noon on November 17, OpenAI CEO Sam Altman connected to a video call from a luxury hotel in Las Vegas. He was in town for the Formula 1 race, among 315,000 visitors, including celebrities like Rihanna and Kylie Minogue.
Mr. Altman, made famous well beyond the tech sector by the success of ChatGPT, had a meeting with Ilya Sutskever, chief scientist of the company specializing in artificial intelligence (AI). But from the start of the call, Mr. Altman sensed something was wrong: Mr. Sutskever was not alone; he was in the company of the three independent members of the OpenAI board of directors.
Unbeknownst to Mr. Altman, Mr. Sutskever and the three administrators had been whispering behind his back for months. They said Mr. Altman had been dishonest and should no longer lead the company leading the race for artificial intelligence. During a meeting the day before, the members of the Board of Directors had unanimously decided to remove Mr. Altman and now they told him the news.
Shaken at being fired from a startup he had co-founded, Mr. Altman asked, “What can I do to help you? » Administrators asked him to support an interim CEO. He assured them he would.
The same day, Mr. Altman would change his mind and declare war on OpenAI’s board of directors.
His firing was the culmination of years of tension within OpenAI, pitting those who worried about the power of AI against those who saw the technology as a unique opportunity for profit and prestige. This schism turned the directors against each other and the general acrimony led to a battle that would shatter the board of directors. When the dust settled, Silicon Valley’s tech elite and corporate interests had triumphed and would now lead the development of AI.
Microsoft threw its weight behind Mr. Altman to preserve its $13 billion investment in OpenAI, followed by many Silicon Valley executives and investors. Some took the fight to Mr. Altman’s $27 million mansion in San Francisco, defending him on social media and protesting the firing in private messages.
Here’s how Mr. Altman, a 38-year-old multimillionaire, would weather that storm, as told by more than 25 people familiar with the events.
Since its creation in 2015, OpenAI has been a ticking time bomb.
Founded in San Francisco by 12 people, including Elon Musk, Mr. Altman and Mr. Sutskever, OpenAI aimed to bring AI to all of humanity. This young techno start-up was non-profit, with a board of directors ensuring that this mission was respected.
Administrators had opposing visions of AI. On the one hand, those who were wary of the dangers of AI; among them, Mr. Musk, who would slam the door in 2018. On the other, Sam Altman and those who were focused on his potential advantages.
In 2019, Mr. Altman – well-connected in Silicon Valley thanks to his startup incubator Y Combinator – was named head of OpenAI. He only owned a handful of shares.
At the start of 2023, three departures reduced OpenAI’s board of directors to six members, including three founders, MM. Altman, Sutskever and Greg Brockman, the president.
Helen Toner of the Center for Security and Emerging Technology at Georgetown University was one of the “effective altruists” who believe AI could one day destroy humanity. AI had long been familiar to Adam D’Angelo as CEO of the question-and-answer website Quora. Tasha McCauley, scientific advisor at the Rand Corporation, had worked on AI governance and taught at Singularity University (in the technological context, the singularity refers to an irreversible point in the future where machines could no longer be controlled by their creators).
This trio shared the fear that AI would become more intelligent than humans.
The launch of ChatGPT in 2022 caused great agitation within the Board of Directors.
This instant success used on a global scale to write love letters and academic works put Mr. Altman in the spotlight. Often seen alongside Satya Nadella, the CEO of Microsoft, at tech events, he has greatly contributed to the notoriety of OpenAI.
But some board members saw ChatGPT’s success as the antithesis of safe AI, two people familiar with their evolving thinking told the New York Times. Soon after, their fears grew with a disagreement with Mr. Altman over appointments to the three vacant board seats.
In September, Mr. Altman discussed an AI chip project in the Middle East with investors, without telling the board. This alarmed its members, said three people familiar with the events.
When Mr. Altman was fired on November 17, a message dropped like a bomb in a WhatsApp thread of about 100 Silicon Valley business leaders, including Meta’s Mark Zuckerberg and Dropbox’s Drew Houston.
“Sam is fired,” the text said.
The discussion thread immediately caught fire. For what ? What did he do ?
At Microsoft, the main shareholder of OpenAI, we asked ourselves the same question. Kevin Scott, CTO, received a call from Mira Murati, his counterpart at OpenAI, announcing Mr. Altman’s dismissal and adding that she would replace him on an interim basis.
Mr. Scott immediately contacted Microsoft headquarters in Redmond, Washington, to have Mr. Nadella removed from a meeting. Stunned, Mr. Nadella called Ms. Murati to ask her the reasons for the dismissal, the OpenAI board of directors having only issued a statement indicating that Mr. Altman had sometimes lacked transparency towards him. Ms. Murati had no further response, said three people familiar with the call.
Mr. Nadella then called Mr. D’Angelo, OpenAI’s lead independent director. What could Sam Altman have done to justify such radical action by the Board? Was there any wrongdoing?
“No,” Mr. D’Angelo replied, sticking to generalities and leaving Mr. Nadella equally speechless.
Shortly after being fired, Sam Altman received a call from a friend, Brian Chesky, CEO of Airbnb, who wanted to know if he could help. Mr. Cheski bombarded him with questions to understand the reasons for the dismissal. Mr. Altman replied that he too was unsure.
At the same time, at OpenAI, employees were demanding clarification. The board connected to a video call with about 15 executives, crammed into a conference room at OpenAI’s San Francisco offices.
Board members said Mr. Altman lied to them, but they could not provide further details for legal reasons.
OpenAI executives then demanded the resignation of the board of directors that same evening, otherwise they would all leave.
That support gave Mr. Altman ammunition. He considered starting a new company, an option that Mr. Chesky and Ron Conway, a Silicon Valley investor and friend, advised against.
Mr. Altman then decided to take back what he felt was his.
The board considered bringing Mr. Altman back on board, but demanded concessions, including bringing in new directors who could control Mr. Altman, and put forward the name of Bret Taylor, former chairman of Twitter. At the same time, the board also sought another interim CEO in case negotiations with Mr. Altman failed.
At 4:30 a.m. on November 20, Mr. D’Angelo was awakened by a panicked phone call from an OpenAI employee. If Mr. D’Angelo did not resign from the board, OpenAI would collapse, the employee claimed.
Mr. D’Angelo realized that the situation had deteriorated considerably in the last few hours: just before midnight, Mr. Nadella posted on X that he was hiring Messrs. Altman and Brockman to lead an AI lab at Microsoft. By morning, more than 700 of OpenAI’s 770 employees had signed a letter saying they could follow Sam Altman to Microsoft if the board didn’t resign.
Among the signatories, one name stood out: Ilya Sutskever had switched sides.
The survival of OpenAI was at stake. Board members had to negotiate.
To break the deadlock, MM. D’Angelo and Altman spoke the next day. Mr. D’Angelo proposed that former Treasury Secretary Lawrence Summers, a Harvard professor, be appointed to the board. Mr. Altman liked the idea.
Mr. Summers spoke with Messrs. D’Angelo, Altman, Nadella and others. Everyone asked him for his opinion on AI and the direction of OpenAI. Mr. Summers wanted to understand the crisis that was shaking the company. He wanted to be sure he could play a mediating role.
Mr. Summers’ appointment convinced Mr. Altman to drop his demand to serve on the board and agree to an independent investigation into his work as head of OpenAI and his dismissal.
Late on November 21, a deal was reached. Sam Altman became CEO again, but without sitting on the board of directors. MM. Summers, D’Angelo and Taylor would be board members, with a non-voting observer spot promised to Microsoft. Ms. Toner and McCauley and Mr. Sutskever would leave the board.
As recently as this week, Mr. Altman and some of his advisers remained furious, demanding that his name be cleared of all suspicion.
“Do you have a plan B to put an end to the rumors about your dismissal? They’re fake and it’s unhealthy,” Mr. Conway texted Mr. Altman.
Altman responded that he was working on it with the OpenAI board: “They really want silence, but I think it’s important to address this quickly. »
