Elon Musk Warns AI Could Kill Humanity

Elon Musk took the stand in the federal lawsuit he brought against OpenAI at the courthouse in Oakland, California, where reporters and jurors were present, per The Guardian and Technology Review. Multiple outlets report Musk warned in testimony that artificial intelligence "could kill us all" (Siècle Digital; Technology Review). Musk testified about a 2015 dinner with Google cofounder Larry Page, saying Page called him "speciesist," and Musk described that exchange as a motivating event in the early formation of OpenAI (Siècle Digital). Technology Review reports Musk told jurors he had provided roughly $38 million of early funding to OpenAI and later described feeling "duped" by company leaders. The Guardian reports Musk seeks removal of Sam Altman and Greg Brockman, unwinding of OpenAI's for-profit conversion, and redirection of $134 billion in damages to the nonprofit arm.

Public reporting from Technology Review identifies a disclosure made during testimony that xAI, Musk's AI company, uses distilled material from OpenAI's models to train its own assistant, including Grok. For practitioners, that admission highlights a common engineering practice-model distillation and transfer learning-that raises concrete IP, reproducibility, and dataset provenance questions when it occurs between large-scale labs. Industry observers should note that model reuse and distillation are technically routine but legally and ethically sensitive when embedded in litigation over corporate forks and nonprofit-to-for-profit conversions.

The courtroom exchanges frame a wider debate linking AI existential-risk rhetoric with very tangible corporate disputes over governance, valuation, and control. Reporting by The Guardian and Le Monde situates the suit in a dispute over OpenAI's 2019 structural changes and the company's trajectory toward a high-value public offering, which some coverage describes as approaching a $1 trillion valuation. Editorial analysis: High-profile testimony invoking existential risk-delivered under oath and amplified by global press-is likely to shape public and regulatory attention on AI oversight, even though such testimony is not itself a technical assessment of system safety.

observers should track:

• •courtroom exhibits and sworn discovery that specify engineering practices, datasets, and model lineage
• •any technical declarations filed by expert witnesses that describe model architectures, training corpora, or distillation methods
• •public statements or disclosures from companies named in the trial that clarify provenance and licensing of training data. Industry reporting has already flagged social-media activity by principals and judicial admonishments about extra-curial commentary, which could affect how parties present technical evidence. Finally, watch for follow-on regulatory or policy attention-coverage of the trial is likely to inform lawmakers and agencies debating governance measures for advanced AI

The trial combines legal claims about corporate governance and remuneration with high-profile testimony about the risks of advanced AI and admissions about intercompany model reuse. For AI practitioners, the case underscores two separate but intersecting issues: the technical realities of model transfer and distillation, and the governance questions that arise when those technical practices are entangled with nonprofit-to-for-profit transitions and large valuations.