Elon Musk testifies xAI 'partly' used OpenAI models to train its own

Elon Musk testified in federal court on Thursday that his artificial intelligence startup, xAI, has used OpenAI's models to improve its own systems. During cross-examination by OpenAI attorney William Savitt, Musk was asked whether xAI had engaged in model distillation with OpenAI. Musk responded that "generally all the AI companies" use such techniques, and when pressed again, said "partly." He added that "it is standard practice to use other AIs to validate your AI." Model distillation is a training method in which one AI model acts as a teacher to pass knowledge to a smaller student model. While commonly used within companies to create cheaper, faster versions of their own systems, it has become increasingly controversial when used across competitors. OpenAI and Anthropic have publicly accused Chinese labs including DeepSeek, Moonshot AI, and MiniMax of distilling their models. Google has described the practice as a method of intellectual property theft that violates its terms of service. OpenAI told a House committee in February that it had taken steps to harden its models against distillation, expressing concern that China could advance autocratic AI by appropriating American innovation. The Trump administration has also weighed in, with White House science and technology policy director Michael Kratsios saying in an April memo that the government would share information with U.S. AI companies about foreign distillation efforts. The admission comes amid Musk's ongoing lawsuit against OpenAI, which he co-founded. Savitt has used the multiday cross-examination to question Musk about his past attempts to take control of OpenAI and about emails and texts from 2017 suggesting he withheld funding and recruited researchers away from the company.