OpenAI must produce Musk testimony

A federal magistrate judge has ordered OpenAI to turn over testimony from its recent courtroom victory over Elon Musk and xAI for use in separate copyright lawsuits, according to Bloomberg Law. The ruling reaches depositions from OpenAI co-founders Sam Altman and Greg Brockman and from Microsoft Corp., tying one of the industry’s highest-profile governance disputes to litigation over how large language models were trained. Bloomberg Law said the order came in federal court as plaintiffs in the copyright cases seek broader discovery into OpenAI’s data sources and internal decision-making. The decision adds to a growing stack of cases that are testing whether AI companies can shield core training practices from outside scrutiny. ### Which testimony has to be produced? Bloomberg Law reported that the material includes depositions from Altman, Brockman and Microsoft gathered during the Musk litigation. The order allows those witnesses and related documents to be used in separate copyright cases against OpenAI, expanding the life of evidence first collected for a different dispute. (news.bloomberglaw.com) The Musk case itself ran in federal court in Oakland, California, and featured about two weeks of testimony from Musk, Altman and other OpenAI figures before closing arguments, Bloomberg Law reported last week. That trial focused on Musk’s claims that OpenAI had strayed from its founding mission after taking billions of dollars from Microsoft and pursuing a for-profit structure. (news.bloomberglaw.com) ### Why do copyright plaintiffs want material from the Musk case? The copyright cases center on how OpenAI trained its models and what records exist about the origin of that material. Bloomberg Law has separately reported that disputes over internal communications about datasets containing pirated books could expose OpenAI to large damages and sanctions risk in copyright litigation. (news.bloomberglaw.com) A May 22 legal analysis published on JD Supra said the “biggest unresolved question” for companies using or training AI tools is whether training on copyrighted material is lawful at all. The article said recent complaints allege AI-generated content can track protected works in plot, sequencing, names and other creative choices, and described copyright risk as central for companies that build or deploy these systems. (news.bloomberglaw.com) ### What are lawyers and academics saying about the broader exposure? Northeastern University reported on May 22 that a growing number of lawsuits seek to hold OpenAI responsible in cases where plaintiffs say ChatGPT played a role in crimes and deaths. The article said experts are split on whether those cases will force industry-wide change, but it described liability questions as moving from theory into active litigation. (jdsupra.com) JD Supra also cited a broader rise in AI litigation. In a separate April analysis reviewing 168 district court cases, the publication said filings accelerated in 2025 and remained elevated in 2026, with disputes concentrating around ownership, authorship and liability. That review was a law-firm analysis rather than a court tally, but it points to the widening range of claims now confronting AI developers. (news.northeastern.edu) ### How is OpenAI responding? JD Journal reported on May 22 that OpenAI has assembled a group of major law firms to handle litigation, regulation and transactions. The publication named Wachtell Lipton, Morrison & Foerster, Latham & Watkins, Cooley and Wilson Sonsini among firms advising the company, and said copyright suits over training data remain one of OpenAI’s biggest legal challenges. (jdsupra.com) JD Journal also reported earlier this week that OpenAI’s win over Musk removed a major obstacle to a future initial public offering plan. That same report said Silicon Valley lawyers are still debating how nonprofit technology organizations should operate after attracting large private investment. ### What comes next in court? The next step is in the copyright cases, where plaintiffs will press for the Musk-case depositions, related documents and additional discovery about training-data provenance, according to Bloomberg Law. (jdjournal.com) The order does not resolve whether training on copyrighted works was lawful, but it gives the plaintiffs access to testimony from some of the people at the center of OpenAI’s development and financing. (jdjournal.com) OpenAI is also facing separate liability suits beyond copyright. Northeastern reported on May 22 that courts are beginning to weigh claims tied to deaths and crimes allegedly linked to ChatGPT, while JD Journal has reported other cases over legal advice and product use. Those proceedings, along with the copyright actions, are the next venues where judges will decide how much of OpenAI’s internal record must be disclosed. (news.northeastern.edu) (news.bloomberglaw.com)