Elon Musk Admits xAI Used OpenAI Models to Train Grok

In a federal court in California on Thursday, Elon Musk acknowledged that his AI firm, xAI, utilized distillation methods on OpenAI models to develop its Grok chatbot. This process involves leveraging publicly available AI models to glean knowledge, facilitating a more cost-effective replication of functionalities. When questioned about xAI's practices, Musk replied "Partly," indicating that such methods are prevalent in the AI industry. This revelation surfaced during Musk's lawsuit against OpenAI, its CEO Sam Altman, and Greg Brockman, claiming they deviated from their nonprofit mission by transitioning to a for-profit model. Musk emphasized the competitive strain, as distillation poses a risk to the investments of major AI firms. He ranked AI leaders, placing Anthropic at the top, followed by OpenAI, Google, and Chinese open-source alternatives, while describing xAI as a smaller entity with a few hundred staff. OpenAI, Anthropic, and Google are reportedly collaborating through the Frontier Model Forum to address distillation efforts, particularly from China. The legal status of distillation is uncertain, as it may breach service agreements rather than specific laws. OpenAI refrained from commenting on Musk's statement.