Musk testifies xAI used model distillation with OpenAI technology in courtroom proceedings

In sworn testimony before a federal court in California, Elon Musk confirmed that his AI startup xAI has used model distillation techniques involving OpenAI's models. When asked directly on the stand whether xAI had distilled OpenAI's technology, Musk initially deflected, noting that "generally all the AI companies" engage in such practices. Upon further questioning, he characterized his answer as "partly" affirmative and elaborated that model distillation is standard practice for companies to validate their own AI systems.

Model distillation refers to a training technique in which a larger, more capable AI model serves as a teacher to transfer knowledge to a smaller, more efficient model. The practice has legitimate applications within single organizations—for instance, frontier AI labs routinely distill their own models to create cheaper, smaller versions for customers. However, the technique is increasingly controversial when applied across company boundaries, where competitors may use it to replicate the performance of rival systems at a fraction of the development cost and time.

The legal and policy status of cross-company model distillation remains unsettled. OpenAI and Anthropic have both raised concerns about the practice, with OpenAI publicly objecting to DeepSeek's apparent distillation of its models, and Anthropic naming DeepSeek, Moonshot, and MiniMax as companies engaging in the technique. Google has similarly moved to prevent what it terms "distillation attacks," treating the practice as a violation of its terms of service. Anthropic's own analysis suggests distillation sits in a gray zone: it is "a widely used and legitimate training method," yet can serve as a vehicle for "illicit" competitive advantage when used to bypass the independent development process.

Musk's courtroom acknowledgment occurs amid the broader Musk v. Altman litigation, which centers on disputes over OpenAI's organizational direction and control. The trial has surfaced various technical and business practices relevant to both companies' competitive positioning in the AI market.