Warning from OpenAI leaders helped lead to Sam Altman’s ouster

Warning from OpenAI leaders helped lead to Sam Altman’s ouster

Senior employees described Altman as psychologically abusive, creating chaos at the AI ​​startup, complaints that were a major factor in the board’s surprise decision to fire the CEO.

Assembled with Sam Altman and Rocket.
(Photo courtesy of Elena Lacey/The Washington Post; TWP; iStock)

This fall, a small number of senior leaders reached out to OpenAI’s board to express their concerns about CEO Sam Altman.

Altman — a revered mentor, prodigious investor in startups and the image of the AI ​​revolution — was psychologically abusive, employees alleged, creating pockets of chaos and delays at the AI ​​startup, according to two people familiar with the board’s proceedings. Thinking spoke on the condition of anonymity to discuss sensitive internal matters. Leaders of the company, a group that includes key figures and people who manage large teams, reported that Altman pitted employees against each other in unhealthy ways, the people said.

Although board members did not use abusive language to describe Altman’s behavior, these complaints echoed their interactions with Altman over the years, and they were already discussing the board’s ability to hold the CEO accountable. Many board members believed Altman lied to them, for example, as part of a campaign to remove board member Helen Toner after she published a research paper critical of OpenAI, the people said.

The new complaints prompted a review of Altman’s behavior, with the board weighing the devotion Altman had cultivated among the company’s factions against the risk that OpenAI would lose key leaders who found interactions with him too toxic. They also took into account reports from several employees who said they feared retaliation from Altman: one of whom told the board The people said Altman was combative after the employee shared critical comments with the CEO and that he belittled the employee on that person’s team.

“Clearly there was a real misunderstanding between me and the board members,” Altman said. Written on X. “For my part, it is very important to learn from this experience and apply those lessons learned as we move forward as a company.”

Complaints about Altman’s alleged behavior, which had not previously been reported, were a major factor in the board’s surprise decision to fire Altman on November 17. Altman’s dismissal was initially seen as a fight over the safe development of artificial intelligence, but it was at least partly true. Driven by his sense that his behavior would make it impossible for the board to supervise the CEO.

Altman was reinstated as CEO five days later, after employees issued a letter signed by a large percentage of OpenAI’s 800 employees, including most senior managers, threatening mass resignations.

“We believe Sam is the best leader for OpenAI,” said company spokeswoman Hannah Wong. “The senior leadership team has unanimously called for Sam’s return as CEO and the Board’s resignation, actions supported by an open letter signed by more than 95% of our employees.”

Now that he’s back at the helm of OpenAI, Altman may find the company less united As the heartwave emojis that greeted his return on social media might suggest.

Some employees said Altman’s camp began undermining the board’s decision shortly after he was removed as CEO. Within hours, letters dismissed the board as illegitimate and denounced Altman’s firing as a coup by OpenAI co-founder and chief scientist Ilya Sutskever, the people said.

On social media, in news reports and on the anonymous app Blind, which requires members to sign up with a work email address to post, people identified as current OpenAI employees described facing intense pressure from their peers to sign a mass resignation letter.

Some OpenAI employees rejected the idea that there was any coercion to sign the letter. “Half the company signed on between 2 and 3 a.m.,” a member of OpenAI’s technical staff, who tweeted under the pseudonym @roon, posted on X. “This is not something that can be achieved through peer pressure.”

Joan Zhang, who works in product at OpenAI, chirp That there was no effect, “The Google Doc crashed, so people texted each other at 2-2:30 a.m. begging people with write access to write their names.”

For longtime employees, there was an additional incentive to sign: Altman’s departure jeopardized an investment deal This would allow them to sell their shares back to OpenAI, and cash out the shares without waiting for the company to go public. Deal Led by Joshua Kushner’s Thrive Capital — the company is valued at about $90 billion, according to a report in the Wall Street Journal, more than triple its $28 billion valuation in April, and would have been threatened by a decline in value caused by the departure chief executive officer.

Board of Directors They expected employees to be upset by Altman’s firing, but were surprised when OpenAI’s management team seemed united in their support for bringing him back, said the people, and a third person familiar with the board’s proceedings, who also spoke on the condition of anonymity to discuss sensitive company matters.

As the company seeks to rebuild its board and smooth things over with Microsoft, its majority shareholder, it has committed to launching an internal investigation into the debacle, which came to light on the Friday before Thanksgiving.

In a post on the company’s blog, the board wrote that Altman was removed as CEO after a review found he had not been “consistently honest in his communications.” The Washington Post previously reported that the board’s vote was caused by a pattern of manipulation and rooted in Altman’s attempts to avoid checks on his power at OpenAI.

Altman himself helped pioneer OpenAI’s unique board structure, according to a person familiar with board procedures at the time. The group has up to nine members and is intended to contain a majority of members who have no financial interest in OpenAI. At the time of Altman’s firing, the number of employees was six: three employees (president and co-founder Greg Brockman, Altman, and Sutskever) and three independent directors (Toner, tech entrepreneur Tasha McCauley, and Quora CEO Adam D’Angelo).

But the lack of specific details about the board’s motives has allowed room for speculation and manipulation. Some of the chatter has focused on Sutskever, who in July was named co-leader of a new AI safety team called “Superalignment,” which aims to ensure that advanced AI systems follow human intentions. His public comments about the potential risks posed by “artificial general intelligence” have paved the way for a narrative of risks for commercial interests.

The pressure on Sutskever to overturn his vote was particularly intense. Less than three days later, he wrote on X’s website that “I deeply regret” participating in the board’s decision. He also added his name to the employee’s resignation letter and pledged to reunite the company.

Altman seemed to agree, quoting Sutskever’s message on the X as well as three red heart emojis.

Sutskever’s future at OpenAI is now uncertain. “We hope to continue our working relationship and discuss how (Sutskever) can continue his work at OpenAI,” Altman wrote in a staff-wide email after his return as CEO.

“There has been a lot of wild and inaccurate reporting about what happened with the board, but the bottom line is that Elijah has publicly stated that Sam is the right person to lead OpenAI, and he is very happy to have him back at the helm.” Sutskever’s attorney, Alex Weingarten, head of Willkie Farr & Gallagher’s litigation practice, wrote in a statement.

On Wednesday morning, Sutskever shared a cryptic post on X about learning several lessons in the past month. “One such lesson is that the phrase ‘the beatings will continue until morale improves’ applies more often than it has any right to,” Sutskever wrote. The tweet was quickly deleted.

You may also like...

Leave a Reply