Elon Musk's legal team called a single AI expert witness at the OpenAI trial: Stuart Russell, a University of California, Berkeley computer science professor who has studied the field for decades. Russell told jurors and Judge Yvonne Gonzalez Rodgers that AGI development carries a stack of risks — cybersecurity threats, misalignment, and a winner-take-all dynamic that pulls labs away from safety work. He co-signed the March 2023 open letter calling for a 6-month pause on frontier AI research. Musk signed that same letter, then launched xAI as a for-profit lab anyway.
Russell's job on the stand was narrow: establish that AI is dangerous enough to take seriously, and that the pursuit of AGI sits in tension with safety. He was not asked to evaluate OpenAI's corporate structure or its specific safety policies, a point OpenAI's attorneys hammered on cross-examination. The judge, after objections from OpenAI, also limited how much of Russell's broader existential-risk argument could be aired in open court.
That limitation matters because Russell's larger concern is exactly the dynamic Musk's lawsuit describes — frontier labs racing each other to AGI, with safety as a casualty. Russell has long argued that governments need to regulate the field more tightly. He just wasn't allowed to spell out the worst-case scenarios for the jury.
Key facts
- 01Stuart Russell, a UC Berkeley computer science professor, was the only AI expert witness called by Elon Musk's legal team at the OpenAI trial.
- 02Russell co-signed the March 2023 open letter calling for a 6-month pause in frontier AI research; Musk signed it too, while founding xAI.
- 03Russell told Judge Yvonne Gonzalez Rodgers that AGI development carries cybersecurity, misalignment, and winner-take-all risks.
- 04OpenAI's attorneys got the judge to limit Russell's testimony, blocking his broader existential-risk argument from open court.
- 05Senator Bernie Sanders has cited AI warnings from Musk, Sam Altman, and Geoffrey Hinton in pushing a moratorium on data center construction.
Musk's broader argument is that OpenAI was set up as a charity focused on AI safety and lost its way chasing revenue. His attorneys are leaning on old emails and founder statements about the need for a public-spirited counterweight to Google DeepMind. Russell's testimony was meant to give that argument technical weight: if AGI really is this dangerous, then a safety-focused nonprofit converting into a profit engine looks worse.
“Russell co-signed the March 2023 open letter calling for a 6-month pause in AI research — the same letter Musk signed while quietly standing up xAI as a for-profit lab.”— Jaeden Schafer
OpenAI's response is straightforward. Russell never audited their safety stack. He never reviewed their internal policies. He's a credentialed academic offering general background, not a witness to anything OpenAI specifically did or failed to do. The cross-examination was designed to put exactly that ceiling on how much his testimony could swing the case.
The contradiction sitting on top of all this is Musk himself. He signed the 2023 pause letter while building xAI, and Grok was later shown — during this same trial — to have been trained using distillation on OpenAI's models, a point covered in earlier testimony from week one. Musk's lawyers want the jury to take AGI fears seriously when those fears indict Sam Altman, and to set them aside when they apply to their own client.
That selective seriousness is now playing out at the policy level too. Senator Bernie Sanders has cited warnings from Musk, Altman, and Geoffrey Hinton in pushing a moratorium on data center construction. Hoden Omar of the Center for Data Innovation pushed back, telling TechCrunch that "it is unclear why the public should discount everything tech billionaires say except when their words can be recruited to fill gaps in a precarious argument."
The financial logic underneath the lawsuit is the part neither side wants to dwell on. OpenAI's founders realized early that competing on frontier models meant a compute bill that only for-profit investors could fund. The same fear of concentrated AGI power that motivated the nonprofit ended up forcing the capital structure that broke the founding team apart. That fracture is what produced this courtroom.
Russell's testimony, even truncated, lands one useful point on the record: the people building the most capable AI systems are also the people warning loudest about them, and they keep building anyway. Whether a jury reads that as hypocrisy, as cognitive dissonance, or as a structural feature of a winner-take-all market is now part of what Judge Gonzalez Rodgers has to manage.
The trial keeps surfacing the awkward truth that AI safety rhetoric and AI commercial strategy have been co-produced by the same handful of people for almost a decade. Musk wants the court to treat the safety rhetoric as binding on OpenAI and irrelevant to xAI. OpenAI wants the opposite. A neutral reading is that both sides used AGI fear to raise money, recruit talent, and seek regulatory advantage, and now want to litigate which usage was sincere. That is a hard question for a jury, and an even harder one for the policy debate this case is quietly setting the terms of.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
