OpenAI trial lays bare rivalries behind start-up’s $852bn rise

On the eve of a trial that will decide the future of OpenAI, Elon Musk threatened to make chief executive Sam Altman and president Greg Brockman “the most hated men in America”.

After two weeks of courtroom clashes, each member of the trio has emerged with a damaged reputation as hours of testimony and a trove of emails and texts laid bare the power struggles and avarice during the early days of the AI start-up.

The trial has painted the most complete picture yet of OpenAI’s messy transition from a scrappy moonshot AI lab into an $852bn commercial juggernaut with almost a billion users.

Musk is seeking to unwind OpenAI’s conversion to a for-profit company and unseat Altman and Brockman. As the case heads into its final week in Oakland, California, with Altman still to testify, the outcome is unclear as unflattering evidence of both sides’ motives stacks up.

Musk’s case hinges on the allegation that Altman and Brockman conspired to “steal a charity” that he endowed with $38mn and a mission to develop artificial general intelligence for the benefit of all humanity.

Brockman was forced this week to defend diary entries that appear to show he wanted to convert the non-profit into a commercial enterprise and get rich, as he revealed he owns a $30bn stake in the company. “It would be nice to be making the billions,” he wrote.

But Musk’s argument came under pressure as details emerged of his own ambitions during a pivotal period when co-founders Musk, Altman, Brockman and Ilya Sutskever grappled over the best course for the start-up.

Musk summoned the group to “the haunted mansion I just bought near San Francisco” in August 2017 to discuss how the start-up could outflank Google while remaining true to its non-profit mission.

Actress Amber Heard, at the time Musk’s partner, served whiskey amid the “carnage” from a party the night before as they discussed the next step for OpenAI, Brockman said on Tuesday. 

The group discussed converting the start-up into a for-profit within weeks, according to notes of the meeting from Shivon Zilis, a former OpenAI board member who is the mother of four of Musk’s estimated 14 children. “Woah, fast!” she added in brackets. 

Messages between Musk and Zilis undermined his case by showing he was happy to convert the charity into a for-profit entity so long as he was the controlling shareholder or allowed to merge it with his car company, Tesla.

“He said he’d experienced what it was like to not have control, and he didn’t like it,” Brockman said. “He said he needed the money for Mars . . . $80bn to create a city there.”

Musk gifted his co-founders each a special edition Tesla Model 3, which Brockman felt was an effort at “buttering us up”.

Later, Brockman and Sutskever told Musk they couldn’t accept his proposals, though Sutskever offered a picture he had drawn of a Tesla in an attempt at goodwill.

“[Musk] stood up and stormed round the table, I actually thought he was going to hit me,” said Brockman. Instead, Musk snatched the Tesla drawing they had brought and stormed out.

Musk quit in early 2018 and has since started his own for-profit AI company. OpenAI says this shows his motives were never charitable and that his lawsuit is driven by frustration that xAI’s Grok chatbot has struggled to compete with ChatGPT.

As Zilis took the stand on Tuesday, Musk unveiled a deal with OpenAI’s fiercest rival Anthropic, giving the group access to xAI’s data centres to train and run its models. 

If the jury finds in Musk’s favour, judge Yvonne Gonzalez Rogers could order OpenAI to revert to a non-profit, which would imperil plans for a trillion-dollar-plus public offering this year.

Musk has also demanded $150bn in damages be returned to OpenAI’s non-profit arm and that Altman and Brockman be removed from the company.

Over the past two weeks, Altman has also come under fire from co-founder and former chief technology officer Mira Murati.

She accused him of undermining her, pitting executives against each other and creating a “chaotic” environment. Tasha McCauley, a former board member involved in briefly ousting Altman from the company in 2023, also described the “toxic culture” that stemmed from his dishonesty. 

Microsoft chief executive Satya Nadella will also be called to testify as the US tech giant is accused of “aiding and abetting” OpenAI. It was an early backer of the start-up, and its $14bn investment has grown into a 27 per cent stake worth more than $200bn.

OpenAI’s conversion to a for-profit, partly to accommodate outside investment from Microsoft, has brought into question whether its mission retains precedence over profit.

David Schizer, a former dean of Columbia Law School who was called as an expert witness, said non-profits were usually “very focused on their donors, what they need and want. It’s very unusual to be this heavily reliant on a private investor.”

Brockman’s digital diary, uncovered during pre-trial discovery, has also been used to paint a portrait of greed at OpenAI that conflicts with its idealistic charter.

One entry during tense negotiations with Musk in September 2017 read: “This is the only chance we have to get out from Elon . . . Financially, what will take me to $1bn?”

In another he wrote “[a] realisation from this is that it’d be wrong to steal the non-profit from him . . . That’d be pretty morally bankrupt. And he’s really not an idiot.”

Musk’s lawyers sought to damage Brockman’s credibility by highlighting that his personal stake in OpenAI has snowballed to $30bn, potentially making him one of the 100 richest people in the world. 

They argued that this showed he had betrayed his fiduciary duties as an original board member of the non-profit and challenged him to donate $29bn back to the charitable arm.

Musk was combative on the stand, accusing OpenAI’s lawyer William Savitt of trying to trip him up with confusing questions.

“They’re designed to trick me essentially . . . I must give a longer answer because any simple answer would be misleading the jury,” Musk said. “If you ask the question, have you stopped beating your wife, [I can’t answer yes or no].” 

Further testimony from Helen Toner, a former OpenAI board member, revealed the manifold cracks in the start-up’s leadership. Toner criticised Murati for being “strikingly unsupportive” and “remarkably passive” after Altman was fired. She also said Murati refused to tell staff that evidence she provided to the board about Altman’s behaviour was a crucial factor in his dismissal. 

“She did not seem to understand, either wilfully or not, that she had a pivotal role to play in legitimising this decision herself,” Toner said. “She was waiting to see which way the wind would blow, and she didn’t realise that she was the wind.”

The discovery process also revealed frantic text messages between Altman and Murati on the day he was fired in November 2023.

“Can you indicate directionally good or bad?” Altman asked Murati about crisis talks with the board. “Directionally very bad . . . Sam this is very bad,” she replied.