Elon Musk testified in a California federal court that xAI partly used OpenAI models to train Grok through distillation. Asked whether the company had used the technique, he first called it a general practice among AI companies and then answered, "Partly." Distillation can involve systematically querying an accessible chatbot or API and using its outputs to train another model. Musk was on the stand in his lawsuit against OpenAI, Sam Altman, and Greg Brockman over the organization's shift toward a for-profit structure. The legal status of distillation remains unclear, although the practice may violate providers' terms of service. OpenAI did not comment on Musk's admission before publication.

The testimony places imitation inside the history of a laboratory founded years after its rival. Training data has produced disputes over books, images, news archives, and code. Model outputs now become another contested resource. A leading system can supply training examples to a competitor that sends enough carefully designed queries. At the same time, OpenAI, Anthropic, and Google reportedly share information through the Frontier Model Forum to combat distillation attempts associated with Chinese firms. Their defense operates through access rules and detection of suspicious mass queries. Musk's answer turns that boundary into a practical question for every lab seeking protection. Providers must decide which pattern of prompts resembles ordinary use and which account should be blocked as a rival's training operation.