Elon Musk attests that XAI trained Grok on OpenAI models

OpenAI and Anthropic have recently been on the warpath against third-party efforts to train new AI models by incentivizing chatbots and publicly accessible APIs, a process known as “distillation.”

That conversation has focused on Chinese companies using distillation to create open-weight models that are roughly equivalent in capacity to American offerings, but are available at a much lower cost. However, technology insiders widely assume that US labs are using these technologies on each other to avoid falling behind competitors.

Now, we know this to be true in at least one case: In federal court in California on Thursday, Elon Musk was asked whether xAI had used distillation techniques in OpenAI models to train Grok, and he confirmed that this is a common practice among AI companies. When asked if that meant “yes,” he said: “Partly.”

Musk is suing OpenAI, CEO Sam Altman, and Greg Brockman, alleging that they violated OpenAI’s original nonprofit mission by converting the entity into a for-profit structure. That trial began this week, and includes testimony from the technology leader.

Musk’s admission is notable because distillation threatens AI giants by undermining the advantage they have built by investing in computing infrastructure. This allows other software makers to create models that are nearly as efficient at a cheap price. There is no small amount of irony here, given the alleged violation of copyright rules by frontier labs in their search for sufficient data to train their models.

It’s no surprise that Musk’s xAI, which launched in 2023, years after OpenAI, is trying to learn from the then-leader in the field. It is not clear that distilling is explicitly illegal, but may even violate the terms of service companies specify for the user of their products.

OpenAI, Anthropic, and Google have reportedly launched an initiative through the Frontier Model Forum to share information on how to combat attempts at distillation from China. These typically involve systematic inquiry into models to understand their inner workings. To stop these efforts, Frontier Labs is working to prevent users from making suspicious group inquiries.

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 technology would soon be beyond the reach of any company besides Google. In response, he ranked the world’s leading AI providers, saying Anthropic took the top spot, followed by OpenAI, Google, and Chinese open source models. He described xAI as a much smaller company with only a few hundred employees.