A MARINE LAYER HUNG OVER OAKLAND on Wednesday, the last day of testimony in the liability phase of Elon Musk’s lawsuit against artificial intelligence company OpenAI and its CEO and president — Sam Altman and Greg Brockman, respectively.
Musk contends that under the leadership of Altman and Brock, the company — founded as a nonprofit charitable enterprise to benefit humanity not individual profit — has been turned into a massive wealth machine for the benefit of insiders and co-defendant Microsoft.
U.S. District Judge Yvonne Gonzalez Rogers, the trial judge, divided the trial into two phases. In the first phase (which is now ending), the jury will determine whether the defendants are liable on Musk’s claims that OpenAI, Altman and Brockman have breached a charitable trust and were unjustly enriched, and if so whether Microsoft aided and abetted the breach.
The jury is technically an advisory jury, and its verdict is not binding on Gonzalez Rogers, though she said she would very likely follow it.
In the second phase, the judge — without input from the jury — will determine what damages or remedies will be imposed for any liability found in the first phase. Musk has requested disgorgement (that is, payment) of up to $134 billion by the defendants.
In addition, Musk has asked for nonmonetary relief that would oust Altman and Brockman and unwind prior transactions as needed so the initial charitable mission can be fulfilled.
Shortly before the trial began, Musk disavowed any interest in obtaining any monetary amounts to be disgorged for himself; he says they should go to what he referred to as “the OpenAI charity,” though how that will work is yet to be explained. Gonzalez Rogers hasn’t yet ruled on the extent that Musk can pursue the non-monetary remedies.
The jury will receive its final instructions from the judge Thursday and is set to begin its deliberations on Monday.
In an unusual twist, the damages phase will begin Monday, even as the jury is deliberating liability. Should the jury return a verdict of no liability before the damage testimony is complete (and assuming Gonzalez Rogers agrees to follow it), there would be no need for a determination of the amount and kind of damages, and presumably the proceedings in that phase would end mid-stride.
Gonzalez Rogers uses a time clock to manage the length of the trial. At the close of pretrial proceedings, she allocated the three parties — Musk, OpenAI (including Altman, Brockman and affiliates) and Microsoft — a fixed number of hours for the presentation of their cases.
Court staff keep track of each party’s use of time like a chess clock. Gonzalez Rogers has left it to the parties to manage use of their allocated time, but she isn’t flexible when it comes to adding time. She said she sees managing the clock as a part of each side’s trial strategy.
Hammering home the details
Most of Wednesday’s witnesses were brief and offered to fill in — or hammer home — points that the jury had already heard.
Microsoft called its chief technology officer and its head of corporate development to back up Microsoft’s central theme that it did a lot of due diligence before investing and its teams found no indication that there were any agreements with Musk about a for-profit subsidiary. Microsoft believed, based on its due diligence, that it was not prohibited from investing in the for-profit.
OpenAI presented three experts to rebut the testimony of Musk’s nonprofit governance law expert David Schizer who previously testified that OpenAI and Altman’s practices and governance were inconsistent with customs and practices of nonprofit corporations.
OpenAI presented Daniel Hemel, a well-credentialed tax lawyer, to testify about the customs of nonprofits that have for-profit subsidiaries. He said he looked at the 100 largest U.S. charities and determined that 92% had for-profit subsidiaries.
The jury will receive its final instructions from the judge Thursday and is set to begin its deliberations on Monday.
He found that it was not exceptional for a for-profit subsidiary to be larger than its nonprofit parent, citing the family trust that operates a school for underprivileged kids in Pennsylvania while owning the far larger $30 billion Hershey candy company.
He also testified that in most cases he examined, the nonprofits had less control over their for-profit subsidiary than OpenAI holds over its for-profit subsidiary.
The second expert was John Coates, a Harvard Law School professor, who has focused on nonprofit governance. He gave his opinion that Schizer’s testimony was repeatedly in error or flatly wrong. He challenged Schizer’s opinion that the OpenAI nonprofit was treated unfairly in transactions with its for-profit and with Microsoft.
Coates said in some cases that Schizer offered opinions completely unsupported by a quantitative analysis. In one case, he said he could not understand how Schizer reached his conclusion.
The third expert, forensic accountant Louis Dudney, traced Musk’s donations to OpenAI and testified that all the proceeds were used in the pursuit of the nonprofit’s mission. His testimony supported OpenAI’s argument that all of Musk’s donations were used for the purposes intended, and therefore Musk hasn’t been wronged.
Chief futurist revisits the past
By far the liveliest testimony came from Joshua Achiam, who today is OpenAI’s “chief futurist.”
Achiam climbed to the stand and greeted the jury with a bright “Hi!” He is years younger than most of the witnesses who have testified so far. He testified that he earned a Ph.D. in artificial intelligence from the University of California, Berkeley.
He joined OpenAI as an intern in September 2017. He became a full-timer in December that year. In his current role as chief futurist, he works on the safe deployment of advanced AI in the future and how to keep the world safe from the technology.
Achiam attended an all-hands meeting in 2018 when Musk stepped down from the board of OpenAI. He said almost all employees (approximately 50 or 60) were in attendance. The format was that Altman would ask Musk a few questions and then open it up to questions from the employees afterward.
Achiam said Musk told employees at the 2018 meeting that he was leaving the board because Tesla planned to accelerate its AI efforts and he expected to recruit engineers who might otherwise work for OpenAI, creating a conflict.
According to Achiam, Musk also said that he was unhappy with OpenAI’s progress in overtaking Google’s DeepMind lab in developing AGI (artificial general intelligence) and wanted to go his own way to move faster.
Achiam said among the questions from the employees were several from the safety staff concerned about the implications if Musk initiated a race to develop AGI.
Achiam joined in and said to Musk that a race for AGI was an “unsafe proposition.” In his opinion, so much was unknown about AGI, and it would be reckless to create it without first making sure it was safe.
He recalled Musk becoming defensive and calling him a “jackass.” Achiam said it wasn’t said in a friendly way but was because Musk “was upset that he had been challenged.”
Standing up to Musk
In Achiam’s telling, the punchline came at the next all-hands meeting when his colleagues presented him with a small trophy with a golden jackass attached to its base. The inscription? “Never stop being a jackass for safety.”
Achiam explained that it was given to him for “for standing up to Musk.”
After the jackass story, Achiam offered broad supportive testimony for OpenAI as well as the leadership of Altman and Brockman. He said they worked unbelievably hard, that the company was deeply committed to safety, and in his view was irretrievably dedicated to its mission.
The cross-examination of Achiam was short and to the point. Achiam admitted that he had held equity in the company.
When asked what it was worth, he said it was substantial but didn’t know exactly because it changed with the valuation of the enterprise.
Alexandra Eynon, one of Musk’s lawyers, asked if it was more than $5 million.
Achiam said, “Yes.”
“Is it more than 10?” Eynon asked.
“Yes.”
While the Jackass trophy provided some humor, the laugh of the day came in the morning before the jury arrived. Steven Molo, Musk’s lead lawyer, and Bill Savitt, OpenAI’s lawyer, were arguing over whether to correct something Molo had said Tuesday when he was examining Altman.
In the context, Molo got hot and said it was “ironic” that the focus was on what he said, when Altman had clearly been lying.
Gonzalez Rogers cut him off sharply, saying, “This entire trial is a giant irony.”