OpenAI and Anthropic have recently intensified their efforts to push back against third-party attempts to develop new AI models by interacting with their publicly available chatbots and APIs—a technique commonly referred to as "distillation."

Much of the discussion has centered on Chinese companies using distillation to create open-weight models that rival the capabilities of U.S. products, but at a significantly lower cost. However, it has been widely assumed among tech workers that American labs also employ these methods on each other to stay competitive. Now, we have confirmation in at least one instance: during testimony in a California federal court on Thursday, Elon Musk was asked whether his company, xAI, has used distillation techniques on OpenAI models to train its Grok system. He acknowledged that it is a common practice among AI firms. When pressed for a direct "yes," he replied, "Partly."

Musk is currently suing OpenAI, CEO Sam Altman, and co-founder Greg Brockman, alleging that they violated the company's original nonprofit mission by transitioning to a for-profit structure. The trial began this week, featuring testimony from the tech mogul. Musk's admission is significant because distillation threatens major AI companies by eroding the competitive edge they have built through heavy investments in computing infrastructure. This technique enables other software developers to create nearly equally capable models at a fraction of the cost. There is a notable irony here, given the bending—and alleged breaking—of copyright rules by frontier labs in their quest for sufficient training data. It comes as no surprise that Musk's xAI, which launched in 2023, years after OpenAI, would seek to learn from the then-industry leader.

It remains unclear whether distillation is explicitly illegal, but it may violate the terms of service that companies set for using their products. OpenAI, Anthropic, and Google have reportedly launched a joint initiative through the Frontier Model Forum to share strategies for countering distillation attempts from China. These efforts typically involve systematically querying models to understand their internal workings. To thwart such attempts, frontier labs are developing measures to prevent users from making suspicious mass queries. OpenAI did not respond to a request for comment on Musk's admission at the time of publication.

Later in his testimony, Musk was asked about a claim he made last summer that xAI would soon surpass every company except Google. In response, he ranked the world's leading AI providers, stating that Anthropic held the top position, followed by OpenAI, Google, and Chinese open-source models. He characterized xAI as a much smaller operation, with only a few hundred employees.