Musk’s lawsuit against OpenAI is situated around its initial week.

      Three days of cross-examination in Oakland revealed some uncomfortable admissions in a $130 billion lawsuit, including that xAI utilizes OpenAI’s models for training. Ultimately, the judge, not the jury, will make a decision.

      Elon Musk took the witness stand in Oakland on Tuesday, recounting a story he has shared over the past two years. He stated that he founded OpenAI in 2015 to prevent advanced artificial intelligence from being controlled by any single entity. He claimed that his former collaborators, Sam Altman and Greg Brockman, had stealthily transformed the lab into a for-profit organization, secured billions from Microsoft, and excluded him. In his view, the lawsuit he filed in 2024 was an attempt to rectify this situation by restoring the original nonprofit structure and reclaiming what he believes was taken from him.

      By the end of the three days, that narrative seemed significantly more contested than it had been on Monday. Musk's case had encountered "some rough spots." His own admissions during cross-examination, the judge’s frequent reminders about the limitations of the dispute, and a series of pre-trial decisions that narrowed the legal claims have collectively made Musk's lawsuit against the world's most valuable AI company appear, at this point, more difficult to win than initially presented.

      The trial commenced on April 28 in the federal courthouse in Oakland, presided over by Judge Yvonne Gonzalez Rogers. A nine-member jury was selected the day prior. Musk, along with his legal team, OpenAI, Altman, Brockman, and Microsoft, are all part of the lawsuit.

      The primary damages claim exceeds $130 billion, although some early reports mentioned $150 billion; regardless, the structural remedies Musk seeks, such as a partial reversal of OpenAI’s for-profit conversion, are arguably the more significant demands.

      The procedural setup is atypical. Despite the jury being convened, its verdict will only be advisory. Judge Gonzalez Rogers will make the final recommendations on liability and remedies, with a ruling expected by mid-May.

      Thus, the trial is less about winning over the jury and more akin to an extended public deposition in front of the judge, who has already streamlined the case before it began by dismissing Musk’s fraud claims and issuing warnings to both parties.

      Over three days, Musk was the first witness. He spent parts of that time on the stand, initially answering questions from his own attorney, followed by rigorous cross-examination from William Savitt, OpenAI’s lead attorney.

      The first uncomfortable moment arose concerning Musk’s commitment to nonprofit principles, a crucial aspect of the case. Savitt presented internal documents and communications that he argued demonstrated Musk had advocated for OpenAI’s transition to a for-profit entity under his control in 2017 and 2018, and had stepped away from the project when that transformation did not occur. “You were never committed to OpenAI being a nonprofit,” Savitt stated. Musk disagreed with this characterization but acknowledged the documents.

      The second awkward moment, which was heard from the gallery, occurred when Musk admitted that xAI, the AI company he founded that created the chatbot Grok, relies on OpenAI’s models, effectively training on outputs from the very system he claims was improperly transitioned to private interests.

      The third issue was procedural. Savitt contended that Musk delayed too long in filing his lawsuit, asserting that key claims were submitted after the relevant statute of limitations had expired. Whether the judge accepts that argument remains to be seen, but the timeline will be part of the official record.

      Even before the opening statements, Judge Gonzalez Rogers had altered the case. Her pre-trial decisions eliminated Musk’s fraud accusations and limited the trial to the more focused question of whether OpenAI violated charitable trust and contract obligations during its restructuring.

      This change makes the case seem less dramatic in presentation, but easier to litigate and arguably places Musk in a more challenging position to win based on his original narrative of betrayal.

      On the third day, Judge Gonzalez Rogers cautioned the attorneys against framing the proceedings as a referendum on AI safety or Altman’s character, as both sides have been inclined to do so. Musk reiterated longstanding concerns about the existential risks posed by AI, an argument that appeared to stray from the legal issue of whether OpenAI’s directors breached their fiduciary responsibilities.

      What’s next? The trial is anticipated to last an additional two to three weeks. Altman is expected to testify, alongside Brockman, Microsoft CEO Satya Nadella, and several of OpenAI’s early engineers. Musk’s expert witnesses, according to court records, will include Berkeley AI researcher Stuart Russell and tax and nonprofit expert David Schizer from Columbia Law School.

      OpenAI is also expected to present its own lineup of governance and AI safety specialists, with reports noting that the defendants aim to highlight Grok’s safety record to the jury.

      Musk could still regain traction. Cross-examinations of his founding partners may lead to their own admissions; the