Elon Musk confirms xAI used OpenAI’s models to train Grok

During a federal court proceeding in California, entrepreneur Elon Musk acknowledged that his artificial intelligence venture, xAI, utilized technology from rival firm OpenAI to enhance its own proprietary software.

Under direct questioning regarding the development of the Grok system, Musk confirmed that his company engaged in a process known as model distillation. This controversial technique involves using a highly advanced, larger artificial intelligence model as a guide to train and refine a smaller, secondary model. While Musk defended the practice by characterizing it as a standard industry method for validating and calibrating artificial intelligence software, his admission brings significant legal and ethical scrutiny to the proprietary boundaries of AI training methodologies. Musk asserted that many prominent organizations within the sector frequently utilize similar techniques to verify their models' performance.

The controversy surrounding model distillation has steadily intensified as leading artificial intelligence developers seek to protect their proprietary research from being copied. Although the technique is widely accepted when organizations use it internally to create cheaper, more efficient versions of their own systems, utilizing a competitor's platform to jumpstart development remains a highly contentious issue. Major technology firms, including Google, OpenAI, and Anthropic, have publicly criticized rival entities for employing this method, occasionally classifying such actions as intellectual property theft that directly violates standard user terms of service. This court testimony highlights the growing friction and lack of clear regulatory guidelines regarding how foundation models are trained and protected across the broader technology industry.