Elon Musk ‘agrees’ to the AI training practice that Anthropic complained to White House against China

May 3, 2026 - 23:23

Elon Musk told a federal courtroom in California this week that his artificial intelligence startup, xAI, has used models from OpenAI to improve its own systems. The statement came as a surprise, given Musk's ongoing legal battles with OpenAI, the company he co-founded and later left. Musk's admission aligns with a controversial AI training practice that Anthropic, another prominent AI firm, recently complained about to the White House regarding China.

The practice involves using outputs from one AI model to train another, often without explicit permission from the original developer. Anthropic had argued that this method, sometimes called "model distillation," could allow foreign competitors, particularly in China, to rapidly copy advanced AI capabilities without investing in the underlying research. The company urged the U.S. government to restrict the practice on national security grounds.

Musk, however, appears to see it differently. In court, he stated that xAI has used OpenAI's models to train its Grok chatbot, though he did not specify the extent of the reliance. This puts Musk at odds with Anthropic's stance, even as he has publicly warned about the dangers of AI falling into the hands of adversaries. Legal experts note that Musk's disclosure could complicate his own lawsuit against OpenAI, where he accuses the company of abandoning its nonprofit mission. The revelation also raises questions about how U.S. AI policy will evolve, especially as companies like xAI and Anthropic push for different regulatory approaches. For now, the courtroom admission has added a new layer to the debate over AI training ethics and national security.