Elon Musk can’t escape his own tweets on stage

Elon Musk appeared in California federal court on Wednesday, claiming that Sam Altman and his co-founders “stole charity.” He left after acknowledging under oath that Tesla is not currently pursuing artificial general intelligence (AGI), which directly contradicts a tweet he posted a few weeks ago.

It was that kind of day for Musk.

The lawsuit he filed challenging OpenAI’s structure alleges that Altman and his other co-founders tricked him into supporting a nonprofit, creating a for-profit arm of Frontier Labs, and controlling the organization.

After hours of testimony from the sometimes acerbic Mr. Musk, it appears the case may come down to how much the jury and Judge Yvonne Gonzalez Rogers distinguish between whether OpenAI’s investors are capping their potential profits.

Musk said that when he co-founded the institute with Altman, Ilya Satskever and Greg Brockman, he trusted them to build AI for humanity, but over time he began to question their motives and ultimately concluded that they were “looting from nonprofits.”

OpenAI lawyer William Savitt tried to complicate the story in cross-examination, trying to show that Musk had supported various efforts to turn OpenAI into a for-profit company so it could raise the funding needed to compete with companies like Google, including building an AI lab into Tesla.

Musk testified that in 2016 he discussed converting the company into a for-profit company, and in 2017 he considered creating a for-profit arm of OpenAI where he would own a majority stake and control the company. When those plans fell apart, he stopped making regular donations to OpenAI, but continued to pay for office space through 2020.

Musk argued that there is a big difference between investors with limited profits and investors with unlimited profits. Microsoft’s early large investments in OpenAI limited the software giant’s profits, but those limits have been rolled back over the years. Musk says these changes ultimately led him to file this lawsuit.

Mr. Savitt did not object to Mr. Musk’s attempt to prove that he was consulted by Mr. Altman and his longtime adviser and mother of his four children, Siobhan Gillis, about subsequent fundraising efforts. Zillis was also a member of OpenAI’s board of directors when it approved some of these transactions.

The cross-examination also extended to Tesla’s AI ambitions. In particular, Musk was asked about Tesla’s efforts to develop competing AI technology, and this wasn’t the first time he realized one of his posts about X was incorrect. After saying that Tesla’s AI research focuses only on autonomous driving, not AGI (a term used to describe AI systems that can perform any intelligent task that humans can do), Musk was asked about a recent post claiming that “Tesla will be one of the companies that develops AGI.” “We are not pursuing AGI right now,” Musk told the court. (Tesla shareholders take note.)

Musk was also asked about a post claiming he invested $100 million in OpenAI, rather than the $38 million that actually changed hands. He argued that his reputation and network compensated for the disparity.

Savitt brought up an email in which Musk supported efforts by Tesla and its brain interface company Neuralink to poach employees from OpenAI while he was still on the company’s board. Another conversation focused on OpenAI’s efforts to hire leaders when they left the board in 2018, including Andrei Karpathy, who left OpenAI to lead Tesla’s self-driving efforts. Mr. Musk was also asked about a conversation in which Mr. Gillis suggested that Mr. Musk recruit Mr. Sutskever to Tesla.

But the most important topic of the day may have been about harm prevention. Part of Musk’s argument is based on the idea that OpenAI’s transition to a traditional company is dangerous to society because it reduces companies’ focus on safety. Savitt then got Musk to admit that all AI companies, including his own, suffer from this risk.

Judge Gonzalez Rogers stopped this series of questioning, but in a statement to his lawyers after the testimony ended, he announced that he would resume questioning with restrictions. When Musk’s lawyers raised questions about ChatGPT’s role in the Tumbler Ridge shooting (an incident in Canada earlier this year in which a man committed murder after having extensive conversations with a chatbot), Musk made it clear that he didn’t want to hear about scandals caused by AI models, but that xAI and OpenAI’s approach to safety was fair.

Musk will return for another round of hostile questioning on Thursday. His family office manager, Jared Birchall, is also expected to testify. AI safety expert Stuart Russell. and Greg Brockman, President of OpenAI.

Correction: An earlier version of this article incorrectly listed details about the Tumbler Ridge shooting due to an editorial error. Updated.