Our Take
Musk's admission that xAI copies OpenAI models while simultaneously suing them for abandoning safety principles exposes the inconsistency at the heart of his case.
Why it matters
The trial could force OpenAI to unwind its corporate restructuring just as it plans to go public this year, potentially reshaping AI's competitive landscape.
Do this week
AI teams: Document your training data sources and model distillation practices before similar disclosure requirements emerge from this case.
Musk admits xAI distills OpenAI models during cross-examination
Elon Musk testified in federal court that his AI company xAI distills OpenAI's models to train its own systems, a revelation that surprised courtroom observers during the first week of his lawsuit against OpenAI. Musk followed up by saying this practice is standard among AI labs, but the admission came while he was simultaneously arguing that OpenAI has abandoned its safety mission.
The case centers on Musk's claim that Sam Altman and OpenAI president Greg Brockman breached the company's charitable trust by converting OpenAI into a for-profit operation after he invested millions in the early days. Musk wants the court to unwind OpenAI's October 2025 restructuring deal with California and Delaware attorneys general that reduced the nonprofit's control over the company.
OpenAI argues Musk knew the company would need a for-profit arm because building AI is expensive. The key dispute is when Musk discovered the alleged misconduct. Under the statute of limitations for charitable trust claims, Musk needed to file within three to four years of learning about it. He claims he only realized in 2022 that OpenAI had abandoned its charitable mission, despite founding the company with Altman and Brockman in 2015.
Text messages revealed during testimony show Musk and Meta's Mark Zuckerberg coordinating to stop OpenAI's restructuring and even considering a joint bid to buy the nonprofit's assets. The judge intervened when Musk's lawyers argued "we could all die as a result of AI," telling them the trial was not about whether AI has damaged humanity.
OpenAI's IPO plans hang in the balance
Even a partial victory for Musk could derail OpenAI's reported plans to go public this year by forcing the company to reverse its corporate structure. The case has become a proxy battle over AI safety governance, with both sides claiming to be better stewards of the technology's risks.
The trial also exposes the level of coordination among Big Tech executives in AI strategy. The revelation that major AI companies routinely distill each other's models raises questions about intellectual property practices across the industry.
Nine jurors will deliver an advisory verdict to guide the judge's decision, though the judge can rule however she chooses. If OpenAI is found liable, the judge will determine appropriate remedies, which could include removing Sam Altman or unwinding the corporate restructuring entirely.
High-profile witnesses still to testify
The trial continues this week with testimony from OpenAI president Greg Brockman, who was observed taking detailed notes during Musk's testimony. Former OpenAI executives including chief scientist Ilya Sutskever and CTO Mira Murati are expected to testify, along with Microsoft CEO Satya Nadella and UC Berkeley computer scientist Stuart Russell on AI safety.
The three-week trial is being closely watched by reporters who must arrive at 6 a.m. to secure one of 30 unreserved courtroom seats. Musk appeared calm and comfortable during testimony, occasionally cracking jokes with lawyers and the judge, though he became flustered during aggressive cross-examination.