Dozens of pages of correspondence between Elon Musk, Sam Altman and other OpenAI founders show how recent conflict over control of the company has deep roots.
SAN FRANCISCO -- ChatGPT maker OpenAI is seen as the leader in developing the artificial intelligence technology captivating the tech industry. It's also known for corporate drama unusual even at a fast-growing start-up, including the firing and reinstatement of its CEO and a recent string of senior departures as it prepares to restructure from a nonprofit organization into a conventional company.
More than 60 pages of emails released in a lawsuit filed by OpenAI co-founder Elon Musk reveal the deep roots of recent tensions inside the company. They show how Musk and other co-founders squabbled from its earliest days over who would control the venture and how it should be funded -- questions that also drove later struggles at OpenAI.
The messages also underscore how a handful of rich men with competing financial interests and ideological goals have controlled a project ostensibly created to work on behalf of all humanity. It has created one of Silicon Valley's most valuable private companies, which recently raised $6.6 billion from investors.
Musk's latest suit, filed late last week, expands a complaint initially filed in February. The billionaire alleges that CEO Sam Altman and Microsoft conspired to overturn OpenAI's original mission of creating AI that is beneficial to humanity and instead chase profits.
OpenAI has said Musk's claims are baseless. A company spokesperson declined to comment.
'Someone other than Google'
OpenAI was founded as a nonprofit in late 2015 after Musk and start-up investor Altman discussed how Google's recent acquisition of AI start-up DeepMind meant the tech giant could establish a monopoly on super-intelligent AI in the future.
"Been thinking a lot about whether it's possible to stop humanity from developing AI," Altman wrote in an email to Musk in May 2015, according to the court filings. "If it's going to happen anyway, it seems like it would be good for someone other than Google to do it first."
Musk agreed, the filings show, becoming a central financial backer of the project and saying in another 2015 email that OpenAI was his "absolute top priority." Musk has said he gave about $50 million to the nonprofit.
But the billionaire, Altman and other key members of OpenAI's founding team were soon at odds over how to secure the additional huge resources the project needed and who should control OpenAI and its future technology, according to the filings.
Late in 2015, just as the project was about to go public, Altman asked Musk whether he could increase the annual pay of everyone who had agreed to join OpenAI by "100-200k," according to the emails, because he heard that Google's AI lab DeepMind was planning to make huge financial offers to "kill" the nascent organization.
Musk responded by offering to call potential employees personally to encourage them, the filings show. Early in 2016, he responded to details of compensation packages that included a $125,000 bonus that could be taken as stock in his rocket company SpaceX by writing "Let's go higher," his court filings show.
"Either we get the best people in the world or we will get whipped by Deepmind."
'Made me feel nauseous'
Despite telling the world that it had $1 billion in committed funding, OpenAI got far less -- it later disclosed that the nonprofit ultimately received donations totaling only about $130 million over the next few years.
On top of paying the six- and seven-figure salaries required to attract top AI researchers, OpenAI also needed to pay for access to high-end computer chips to power its projects.
In September 2016, according to the court filings, Altman sent Musk a draft agreement between OpenAI and Microsoft. It offered the AI start-up $60 million worth of cloud computing access for just $10 million, allowing it to train its AI models on Microsoft's computer chips. In apparent exchange for the discount, the agreement asked OpenAI to help "evangelize" Microsoft's products, which Musk opposed, according to the email exchange.
"This actually made me feel nauseous. It sucks and is exactly what I would expect from them," Musk wrote back to Altman, according to the court filings. But other emails released in the lawsuit show the billionaire was soon accused by his co-founders of wanting to compromise OpenAI's independence.
In 2017, OpenAI's founders realized that their mission would require billions of dollars annually that only a for-profit entity could secure, according to a company response to Musk's suit earlier this year.
An Aug. 28, 2017, email to Musk released in the court filings from Shivon Zilis, a Tesla executive and OpenAI adviser who later became mother to three of Musk's children, summarizes a conversation she had with two other key co-founders.
Greg Brockman, OpenAI's first chief technology officer and an experienced entrepreneur, and Ilya Sutskever, its top researcher who had been hired away from Google, had concerns about the company's financial future, Zilis wrote.
She said the pair insisted on a provision that would ensure no single person controlled OpenAI if it invented what they called AGI, or artificial general intelligence, a hypothetical future form of AI as smart and flexible -- or smarter -- than humans. They also wanted Musk to spend more time working on OpenAI, her summary said.
"This is very annoying," Musk replied, according to the court filings, telling Zilis to encourage Brockman and Sutskever to leave OpenAI and start their own company. "I've had enough."
'You could become a dictator'
A month later, according to the filings, Sutskever and Brockman sent an impassioned email to Musk and Altman. It laid out concerns that either of the two moguls could exploit changes to OpenAI's structure to amass more power for themselves.
"You stated that you don't want to control the final AGI, but during this negotiation, you've shown to us that absolute control is extremely important to you," they wrote in a section of the exchange addressed to Musk. "It is a bad idea to create a structure where you could become a dictator if you chose to."
To Altman, they wrote, "we haven't been able to fully trust your judgments throughout this process," according to the filings. "... Is AGI truly your primary motivation? How does it connect to your political goals?"
Musk responded to the email with an ultimatum of his own, according to his lawsuit. "I will no longer fund OpenAI until you have made a firm commitment to stay or I'm just being a fool who is essentially providing free funding for you to create a start-up. Discussions are over."
Subsequent emails released in Musk's filing show Zilis telling him that Altman, Brockman and Sutskever were happy for OpenAI to continue as a nonprofit -- but by the next year, the founders started arguing again over how to fund the expensive project.
Without major corporate funding, OpenAI was reliant on donations. In January 2018, Altman proposed that OpenAI create its own cryptocurrency that it could sell to raise more funds, according to the filings.
Musk shot down the idea, the released emails show, writing that it would "result in a massive loss of credibility for OpenAI" and that considering it had been, "in my opinion, an unwise diversion." He also warned that "OpenAI is on a path of certain failure relative to Google."
Days later, Musk proposed solving the funding problem by subsuming OpenAI into his own empire, as part of Tesla.
"Tesla is the only path that could even hope to hold a candle to Google," Musk said in an email to Brockman and Sutskever in the court filings. "Even then, the probability of being a counterweight to Google is small. It just isn't zero."
'I have no financial interest'
Musk officially left OpenAI soon after, but Altman and Zilis continued to send him emails about the company, according to the court filings.
In April 2018, they show that Altman asked Musk for his thoughts on releasing OpenAI's "charter," a document that included commitments like "our primary fiduciary duty is to humanity," and to make sure that if super-intelligent AI was developed, it would benefit everyone. Musk responded: "Sounds fine."
Nearly a year later, in March 2019, Altman became OpenAI's full-time chief executive. The same month, the nonprofit announced that it had created a "capped-profit" subsidiary.
It could provide a return to investors up to a total return of 100 times their original stake -- a level few investments ever reach -- after which additional profits would go to OpenAI's original nonprofit board, which remained tasked with spreading the benefits of AI to all.
The last email exchange in the filing takes place the day the capped-profit structure was announced, March 11, 2019. Musk sent Altman the headline from a news report on the announcement that mentioned his role as co-founder of the company. "Please be explicit that I have no financial interest in the for-profit arm of OpenAI," Musk wrote to Altman. "On it," he responded.
Despite Musk's and Altman's early fears, Google did not end up dominating AI technology. Instead, the reformulated OpenAI won major backing from Microsoft and became the company to beat after the 2022 release of ChatGPT triggered a flood of excitement about AI.
Musk responded by launching his second new AI company, xAI -- this time wholly under his control. He and Altman have not stopped sparring over the best path forward for AI technology.
The day Musk filed his updated lawsuit with the trove of emails last week, Altman taunted his former collaborator on the social network X. He took aim at xAI's chatbot Grok, which Musk has positioned as a less politically biased alternative to ChatGPT.
In his X post, Altman shared screenshots that appeared to show Grok favoring Vice President Kamala Harris over Donald Trump, now the president-elect. "Which one is supposed to be the left-wing propaganda machine again?" he wrote.
Nitasha Tiku contributed to this report.