ELON MUSK’S LAWSUIT against Sam Altman and OpenAI — now in its fourth day of testimony — continued Monday in the federal courthouse in Oakland with the testimony of Greg Brockman, one of the original founders of OpenAI and a defendant in the case.
Musk asserts that Altman and Brockman, aided and abetted by co-defendant Microsoft, betrayed the charitable mission of OpenAI to enrich themselves and Silicon Valley insiders.
Steven Molo, Musk’s lead lawyer, called Brockman to the stand. Brockman is president of OpenAI and an important witness for Musk, in part because of a personal journal that he maintained during some of the key events at issue in the lawsuit.
Before the jury was brought into the courtroom, the judge heard arguments from the lawyers about the admissibility of an alleged text on April 25 — a few days before the trial began — from Musk to Brockman feeling out a possible settlement. Brockman allegedly proposed a “walkaway” resolution, where each side would drop their claims.
Musk allegedly wrote back, “By the end of this week, you and Sam will be the most hated men in America. If you insist, so it will be.”
The defendants contended that the message was “menacing” and a “threat” and asked for permission to introduce the information in court by way of Brockman’s testimony. U.S. District Judge Yvonne Gonzalez Rogers denied the request without hearing argument on grounds that the defendants should have introduced the message when Musk was on the stand, not “by the backdoor” through Brockman.
Greg Brockman testimony
Molo’s examination of Brockman began with a key sequence of questions about Brockman’s personal circumstances.
Under Molo’s pointed questioning, Brockman agreed that he was part of the OpenAI founding group with Altman and Musk, that he was there when a for-profit subsidiary was created, that he was given an equity stake in the for-profit, that he did not contribute any money to get that stake, and that the stake is today worth close to $30 billion.
The $30 billion became an ongoing theme of Molo’s questioning. Molo hammered Brockman, prodding him to admit that growing a $30 billion fortune was far outside OpenAI’s mission to develop artificial intelligence as a nonprofit and violated Brockman’s fiduciary duties to the nonprofit.
Molo turned to the period in the fall of 2017 that has already been the subject of extensive testimony in the trial. At that point, Brockman said, the founders had a series of “intense” discussions among themselves about ways to raise more money for the development of AGI, or artificial general intelligence. AGI is AI that can “match or exceed human capabilities” over a very broad range of activities. One of the ideas was to create a for-profit corporation that would be able to attract investment dollars and give employees a stake in the future success of the startup.
In those discussions, Brockman and co-founder Ilya Sutskever proposed an equity structure that was ultimately not implemented. Under the proposal, Musk would have held a bare majority, while Brockman and Sutskever would each have had 11%.
Molo used the situation to lay the seed that Brockman was already seeking a way to earn riches. He showcased entries from Brockman’s personal journal in August 2017 in which Brockman was considering which path — profit or nonprofit — would be best. Brockman wrote, “Financially, what will take me to 1B?”
Molo used Brockman’s journal to suggest that, even as Brockman and Altman were committing to the nonprofit structure to Musk, he and Altman were considering how they could get out of that commitment.
Molo prodded Brockman to admit that growing a $30 billion fortune was far outside OpenAI’s mission and violated Brockman’s fiduciary duties to the nonprofit.
He called out reflections in the journal that suggested Brockman understood that if they were to create a for-profit subsidiary for the nonprofit OpenAI, it would go back on assurances given to Musk and would make him and Altman “liars.”
Brockman observed in his journal: “Can’t see us turning this into a for-profit without a very nasty fight. … His [Musk’s] story will correctly be that we weren’t honest with him in the end about still wanting to do the for-profit just without him.”
Gonzalez Rogers cited the testimony in her Jan. 15 opinion denying the defendants’ motion for summary judgment. Summary judgment is the legal term for a court resolution of a claim without the need for trial because the facts are not in dispute.
Molo on a roll
Molo’s strongest moment came when Brockman testified about his “path to 1B” journal entry. Brockman said he meant that a billion dollars would be an outcome that would make all the “blood, sweat and tears” worthwhile.
Molo noted that Brockman in fact got $30 billion, not $1 billion. He got Brockman to confirm that he owed fiduciary duties to the nonprofit to carry out its mission. Then he asked Brockman if he had given back the extra $29 billion.
Brockman struggled to frame a direct response but finally conceded that he had not.
Molo then walked through several transactions between OpenAI and third-party companies in which Brockman holds an interest. He charged that Brockman had a conflict of interest in those transactions and attempted to use them to support the theme that Brockman was out to enrich himself.
Brockman was generally calm and precise in his answers, though at one point he said, “I still don’t understand what I am being sued for.” Molo ran with that ball, as Brockman tried to explain what he meant. Molo used the moment to illustrate the disconnect between the stated mission of OpenAI and Brockman’s concept of what was permissible.
Altman sat quietly in court listening to Brockman’s testimony.
After Molo finished his examination, Brockman’s attorney, Sarah Eddy, walked him through testimony. Brockman became visibly more energetic when discussing the months surrounding OpenAI’s founding.
During his testimony last week, Musk claimed a central role in the recruitment of Sutskever, then a leading AI scientist at Google Deep Mind, to the OpenAI founding team. (Sutskever is no longer at OpenAI and was not named as a defendant.)
Brockman told a different story, underscoring the dispute over Musk’s role in OpenAI’s early days.
He said he was in the middle of those efforts and told a story about his involvement in the efforts to recruit Sutskever.
According to Brockman, he managed the difficult task of getting Sutskever to finally agree. Sutskever would commit to leaving Google and then “waver.” Brockman said he had several conversations with Sutskever during this period. As the deadline for announcing the OpenAI launch approached, Brockman pressed him for a commitment and, in a last-minute text message, got a yes. Sustskever would join the Open AI team.
Brockman repeatedly defended OpenAI’s commitment to its mission and work that would benefit society, pointing to the nonprofit’s work on “AI resilience” and public health. OpenAI is in a partnership to use AI to accelerate Alzheimer’s research, according to his testimony.
Brockman repeated his central point: the nonprofit today has a substantial stake in the for-profit, a stake worth as much as $200 billion, making it “one of the most resourced” charities in the world, able to do a great amount of good. In his view that represents extraordinary success in fulfilling its mission.
Brockman will return for further testimony Tuesday.
Stuart Russell testimony
Also on Monday, Stuart Russell, professor of computer science at UC Berkeley, appeared as an expert witness for Musk. Russell is a prominent figure in the world of artificial intelligence, known for his assessments of the risks of uncontrolled AI development.
Under questioning from Musk’s attorney, Russell outlined the risks he sees from AI, including threats on a “societal scale,” such as very high rates of unemployment, the rapid spread of disinformation, and allowing “people with evil intent” to magnify their efforts with powerful AI tools.
Russell testified that those risks are magnified as companies like OpenAI race to create AGI.
The potential economic value of AGI, Russell testified, creates enormous financial incentives to be the “first mover.” He spoke of a “winner-take-all” risk, where “a handful of first-mover companies would come to “control a majority of economic activity on the planet.” Those huge economic stakes “may lead to safety being a secondary consideration.”
Russell said that the question of how to make AI systems safe was “an unsolved scientific problem.” Until it is solved and guardrails put in place, “each company feels that it needs to be in this race,” he said.
Under cross examination, Robert Kry, one of Musk’s attorneys, told Russell that Tesla, one of Musk’s companies, was engaged in an effort to develop AGI. Kry asked if the financial incentives “would apply to Mr. Musk as well.”
Russell said he did not know if Tesla was developing AGI but said, “If that hypothesis is correct,” Russell said, “then, yes.”
On day two of the trial, Musk testified that he believed AGI would be achieved “as soon as next year.”