Elon Musk admits xAI used OpenAI technology to train its Grok model during trial.

Elon Musk testified during his ongoing lawsuit against OpenAI that his artificial intelligence company xAI trained its Grok model using OpenAI's technology, a practice he characterized as standard in the AI industry. The admission, made under oath during the trial, has become central to disputes over model development practices as the case continues into its fourth day, with Musk concluding his testimony this week.

The revelation about Grok's training methods emerged as Musk faced intense cross-examination from OpenAI's lead attorney, William Savitt. According to reporting on the proceedings, when questioned about whether xAI had used OpenAI's models, Musk argued that such "distillation"—the process of training smaller models on outputs from larger ones—represents common practice among competing artificial intelligence laboratories. This explanation marked a notable moment in the trial, as the practice has become increasingly contentious among frontier AI companies seeking to protect their proprietary models from smaller competitors attempting to replicate their capabilities.

The trial itself has exposed significant tensions between the two parties over the evolution of OpenAI from its nonprofit origins. Musk, who provided early funding he claims totaled $38 million, has sued OpenAI and CEO Sam Altman, alleging breach of contract and the "looting" of the nonprofit organization. His legal team argues that his early backing represents between 50% and 75% of the nonprofit's stake in the company, though OpenAI's own statements put Musk's total contributions at less than $45 million compared to more than $90 million from other donors. Musk has asserted he believed the organization would maintain its nonprofit status indefinitely, but OpenAI transitioned to a hybrid "capped profit" model in 2019 and now operates with a reported valuation around $800 billion.

During his three days of testimony, Musk presented an emotional account of his relationship with OpenAI, describing his support as initially "enthusiastic" before shifting to "uncertainty" and ultimately to conviction that the organization's leadership had "stolen from a nonprofit." He acknowledged in court that he transferred $5 million quarterly to OpenAI and paid an additional $3 million annually in office rent, stating he did so even while harboring concerns because he trusted co-founder Sam Altman's assurances about maintaining nonprofit principles. However, cross-examination by Savitt focused on contradictions in Musk's account, using his own emails and messages—including communications from Shivon Zilis, who acted as an intermediary between Musk and OpenAI leadership—to suggest the lawsuit represents a competitive grievance rather than a principled stand for charity.

Musk explicitly called himself "a fool for giving free money" during his testimony, describing the irony that his $38 million in early financing helped establish what became an $800 billion company from which he received no equity stake. When pressed about Microsoft's involvement, Musk claimed the software giant had reacted "quite negatively" to the prospect of his investment, suggesting that Microsoft's objection had contributed to the structural decisions that eventually led to his exclusion from the company's profitable operations.

The case continues as the industry grapples with broader questions about model training ethics and competitive practices. The admission about Grok's training underscores how smaller AI companies rely on larger competitors' publicly available outputs to develop their own systems, a practice that sits in a legal and ethical gray area. Meanwhile, OpenAI's leadership has moved forward with confidence in the company's market position, with Chief Financial Officer Sarah Friar recently stating the company is meeting financial objectives and experiencing "a vertical wall of demand" for its products, indicating the lawsuit has not dampened investor or customer interest.