In the U.S. federal lawsuit against OpenAI, Musk admitted that xAI, a company he founded, had utilized OpenAI models to train its chatbot Grok using 'distillation' techniques, bringing this industry-wide unspoken practice to light. OpenAI and Anthropic have recently frequently accused Chinese companies of training models through distillation techniques, but Musk's admission reveals that similar practices also exist among U.S.-based labs. The so-called 'distillation' technique involves training new models by intensively querying publicly accessible chatbots and APIs. This approach is seen as a threat to large AI companies, which invest heavily in building computational infrastructure, as distillation could allow other developers to train models with similar capabilities at a lower cost. Currently, the legal boundaries of distillation remain unclear, being more constrained by companies' terms of service rather than statutory law.