In a federal courtroom in California on Thursday, Elon Musk testified that his own AI startup, xAI, has used OpenAI\u2019s models to improve its own.<\/p>\n
The matter at question is model distillation, a common industry practice by which one larger AI model acts as a \u201cteacher\u201d of sorts to pass on knowledge to a smaller AI model, the \u201cstudent.\u201d Although it\u2019s often used legitimately within companies using one of their own AI models to train another, it\u2019s also a practice that\u2019s sometimes used by smaller AI labs to try to get their models to mimic the performance of a larger competitor\u2019s model.<\/p>\n
Asked on the stand whether he knew what model distillation was, Musk said it\u2019s to use one AI model to train another. When asked whether xAI has distilled OpenAI\u2019s technology, Musk seemed to avoid the question, saying that \u201cgenerally all the AI companies\u201d do such a thing. And when asked if that was a yes, he said, \u201cPartly.\u201d<\/p>\n
When pressed, Musk said, \u201cIt is standard practice to use other AIs to validate your AI.\u201d<\/p>\n
Model distillation has been on the rise and has incited more controversy among AI labs, in recent years, since the lines for what\u2019s legal \u2014 and what violates a company\u2019s certain terms or policies \u2014 often fall within a gray area. Companies like OpenAI and Anthropic have accused Chinese firms of distilling their models, with OpenAI publicly stating<\/a> its concerns about DeepSeek, and Anthropic specifically naming DeepSeek, Moonshot, and MiniMax. Google, also, has taken steps<\/a> to try to prevent what it calls \u201cdistillation attacks,\u201d or \u201ca method of intellectual property theft that violates Google\u2019s terms of service.\u201d<\/p>\n