Even artificial intelligence would have struggled to write a story with as many twists and turns as the one that has unfolded at OpenAI over the past few days.
On Friday (17 November), Sam Altman, who co-founded the AI research company in 2015 and served as its CEO from 2019, was sacked by the board. After five days of boardroom jostling, firings and hirings, and an attempted staff coup, we are back to square one, with Altman returning to the helm.
But how did this all happen – and why?
How OpenAI’s board governance set the stage for chaos
To truly understand how this was able to happen, we must go back to 2015, when OpenAI was founded. The company, which is responsible for developing ChatGPT, began as a non-profit, with a core mission to ensure that the development of artificial general intelligence “benefits all of humanity”.
While this altruistic mission was admirable, it hit a snag as the cost of the computational power and talent needed to keep advancing its AI projects began outstripping donations.
A solution to this was devised in 2019 when the organisation restructured from being purely non-profit to a “capped profit” company. The restructure established a new for-profit subsidiary, which would be responsible for raising the capital required for OpenAI to continue to advance its AI models.
This new subsidiary would still be governed by a non-profit board, which would hold responsibility for ensuring OpenAI adhered to its founding mission – to build AI that was safe and for the benefit of all.
The board was split evenly between OpenAI employees (Altman, OpenAI chair Greg Brockman and its chief scientist Ilya Sutskever) and external directors (Quora CEO Adam D’Angelo, robotics engineer and entrepreneur Tasha McCauley, and Helen Toner, director of strategy at Georgetown University’s Center for Security and Emerging Technology).
It was determined that the six directors’ fiduciary duty would be to “humanity, not OpenAI investors” and, if it feared the company was straying from this mission, it retained the power to remove the CEO.
How the chaos at OpenAI unfolded
A timeline of events
Friday
17 Nov
Saturday
18 Nov
Sunday
19 Nov
Monday
20 Nov
Tuesday
21 Nov
Friday
17 Nov
Saturday
18 Nov
Sunday
19 Nov
Monday
20 Nov
Tuesday
21 Nov
Friday
17 November
Saturday
18 November
Sunday
19 November
Monday
20 November
Tuesday
21 November
Friday
17 November
Saturday
18 November
Sunday
19 November
Monday
20 November
Tuesday
21 November
17 November
The OpenAI board fires CEO and co-founder Sam Altman. OpenAI CTO Mira Murati is handed the interim CEO role
18 November
Investors, including Thrive Capital, Sequoia Capital, Tiger Global, and Microsoft, push for Altman’s return
19 November
Board member and chief scientist Ilya Sutskever says Altman won’t be returning to OpenAI.
More staffers quit
20 November
Microsoft hires Altman, Brockman and other ex-OpenAI staff to head up a new “advanced AI research” team
20 November
Former Twitch boss Emmett Shear is appointed as the new interim CEO of OpenAI, replacing Murati. Shear commits to investigate Altman’s dismissal
20 November
Almost all of OpenAI’s 770 employees sign a letter threatening to quit if the board doesn’t resign
20 November
Tech companies, including Salesforce, attempt to capitalise on the turmoil by trying to poach OpenAI’s talent
21 November
OpenAI reaches an agreement in principle for Altman to return as CEO and the board is overhauled, with Bret Taylor and Larry Summers joining
Why was Sam Altman fired as OpenAI CEO?
On Friday afternoon, Altman was told by Sutskever and the three non-employee board members that they had reached a majority decision to fire him. Brockman, the remaining board member, was also informed that he would lose his position on the board but would remain an OpenAI employee.
The board did not fully divulge its reason for reaching the decision, stating only that Altman “was not consistently candid in his communications” and that it “no longer has confidence in his ability to continue leading OpenAI”. Mira Murati, the company’s chief technology officer, was chosen to step in as interim CEO.
Although the true reasoning for Altman’s dismissal remains murky, The New York Times reported that relationships between Altman and Toner had soured after one of her research papers appeared to raise questions about OpenAI’s safety practices. Altman’s efforts to rapidly commercialise OpenAI’s models were also ringing alarm bells.
These concerns feed into a wider rift within the AI community between the optimists, who think that AI development needs to accelerate, and ‘doomers’, who are wary of the speed of advancement and fear the technology poses an existential threat to humanity. It is claimed that Sutskever erred more on the side of caution in this debate, leading to his decision to side with the other three board members.
Altman was also known to be exploring projects away from OpenAI – including the creation of an AI device with former Apple chief designer Jony Ive. It has been speculated that this may have also played a role in the boards’ decision-making.
OpenAI appoints Emmett Shear and Microsoft poaches Altman
Investors, including Microsoft, which reportedly holds a 49% stake in the company, immediately pushed for Altman to be reinstated as CEO. But discussions quickly reached an impasse.
On Monday morning, Twitch co-founder Emmett Shear was appointed as the new interim CEO of OpenAI, becoming the company’s third chief executive in as many days. Shear’s opinions on AI appear to align more closely with that of the board and he has shared his belief that AI development should be slowed down by between 10% and 20%.
However, Shear did clarify in a tweet that: “The board did not remove Sam over any specific disagreement on safety, their reasoning was completely different from that.”
At almost the same time, Microsoft CEO Satya Nadella announced that Altman, Brockman and other OpenAI staff would be joining his company to lead a new “advanced AI research team”.
However, Nadella appeared to leave the door open for Altman’s return. When asked by Bloomberg TV on Monday evening who would be OpenAI’s CEO in 24 hours time, he said: “I will leave it with OpenAI and its board.”
What was the response to Altman’s sacking?
Altman was a popular figure within OpenAI and the decision of the board to oust him, without giving a clear reason why, caused consternation within the company’s ranks.
Brockman quickly announced his resignation in protest of the decision. Reports emerged on Sunday that “dozens” of OpenAI staff were following suit.
Murati, the chief operating officer and the strategy officer were among a number of OpenAI employees to tweet “OpenAI is nothing without its people”, in an expression of solidarity with Altman and Brockman.
In a shock twist, Sutskever also backtracked on his decision, stating that he deeply regretted his role in the board’s actions and “never intended to harm OpenAI”.
He was among more than 500 OpenAI staff who initially signed an open letter threatening to leave unless the board resigned. “Your conduct has made it clear you did not have the competence to oversee OpenAI,” it read.
By the end of the day, almost all of the company’s 770 employees had attached their name to the letter.
Altman is reinstated as OpenAI CEO
With pressure mounting on the board, staff threatening to quit en masse and Sutskever having changed his mind, it appeared to only be a matter of time before Altman was brought back into the fold.
On Wednesday, OpenAI shared that it had reached an “agreement in principle” for Altman to return as CEO, alongside a board restructure. Brockman has also returned to the company.
Shear, who served as interim CEO for only three days, has said he is “deeply pleased” with the outcome. He had committed to hiring an independent investigator to generate a full report on Altman’s dismissal and was reportedly prepared to quit if the board could not provide him with a full explanation for its original decision.
With Altman back in place as OpenAI CEO, it may appear that all the tumult has amounted to nothing. However, there have been some key changes at board level. Bret Taylor, the former co-CEO of Salesforce, has been brought in as chair, while former US secretary of the treasury Larry Summers is also a new appointee. D’Angelo is the only member of the original board to retain their position.
Nadella has also expressed his desire to be kept in the loop, due to Microsoft’s substantial investment in OpenAI, saying that he didn’t want any more “surprises”. This could see a Microsoft representative join the board at a later point.
After a hectic few days, OpenAI and its employees will be hoping that this is the end of the drama. But there could yet be another page to this story, while the fallout and its impact on the progress of artificial general intelligence is likely to continue for some time.
Even artificial intelligence would have struggled to write a story with as many twists and turns as the one that has unfolded at OpenAI over the past few days.
On Friday (17 November), Sam Altman, who co-founded the AI research company in 2015 and served as its CEO from 2019, was sacked by the board. After five days of boardroom jostling, firings and hirings, and an attempted staff coup, we are back to square one, with Altman returning to the helm.
But how did this all happen – and why?
