Elon Musk testified in court that his artificial intelligence company, xAI, partly used models from OpenAI to train its Grok chatbot. The admission confirms the use of a technique called distillation, a lower-cost method for replicating model behavior. The disclosure came during Musk’s ongoing lawsuit against OpenAI, which alleges the company abandoned its original nonprofit mission.
Elon Musk said his artificial intelligence company xAI partly used OpenAI models while training its Grok chatbot. The admission was made during testimony in a California federal court where Musk is suing OpenAI, CEO Sam Altman, and co-founder Greg Brockman.
Musk reportedly said the answer was “partly,” and described the approach as a broader industry practice. The method, known as distillation, allows companies to replicate model behavior at a lower cost.
Distillation involves training a new AI system by querying an existing model and using those outputs. In February, Anthropic accused several Chinese AI developers of using fraudulent accounts to extract responses from its Claude chatbot to train competing systems.
Earlier this month, the White House warned of “industrial-scale” campaigns using proxy accounts to replicate U.S. AI capabilities. Musk’s testimony indicates the method is also being used by U.S.-based AI companies.
The legal boundaries of this practice remain unclear. Distillation is not explicitly illegal but can raise questions about violations of platform rules or API terms of use.
Musk co-founded OpenAI in 2015 as a nonprofit before leaving in 2018. His lawsuit centers on claims that the company moved away from its original mission.
