Elon Musk Admits xAI Has Used OpenAI's Models in AI Training During Court Testimony

Summary

During cross-examination in federal court on Thursday, Elon Musk acknowledged that xAI has used model distillation—a technique where one AI model trains another—with OpenAI's models. When pressed by OpenAI's attorney William Savitt on whether xAI had engaged in distillation with OpenAI, Musk initially deflected by noting that "all the AI companies" practice this technique, before clarifying that using other AI systems to validate your own is "standard practice."

Model distillation is a well-known technique in machine learning where a smaller, more efficient model learns to mimic the behavior of a larger, more capable model. While the practice has legitimate uses for testing and validation, it also enables faster and cheaper deployment of similar capabilities. OpenAI has been aggressively trying to prevent competitors—particularly Chinese AI labs—from distilling its models, having taken technical measures to "harden" its systems against the practice.

Musk's testimony comes as part of his broader legal battle with OpenAI and reflects escalating tensions in the AI industry over competitive practices. The U.S. government has also waded into the issue, with the White House's office of science and technology policy announcing plans to share information with American AI companies about foreign distillation attempts. Meanwhile, other AI labs have begun restricting access entirely: Anthropic blocked both OpenAI and xAI from using its models for coding in recent months.