AI Briefing
Sign in

Nvidia CEO Jensen Huang Defends AI Distillation as Competition Amid US-China Tensions

·2026.09.28 22:55

Key point

Huang's comments contradict Treasury Secretary Scott Bessent, who recently labeled distillation 'theft' and threatened sanctions against overseas companies using it.

1 / 4

Details

Nvidia CEO Jensen Huang publicly rejected the characterization of AI model distillation as "theft," arguing instead that training or learning from competitors’ products constitutes legitimate competition. Speaking on CNBC’s Squawk Box, Huang stated, "You’re allowed to test somebody else’s products all you want," acknowledging that while he would prefer competitors not learn from Nvidia’s products, "competition makes everything better."

US-China AI Flashpoint

The dispute over distillation—training models on the outputs of other models—has become a central tension in the US-China AI race. US officials, including Treasury Secretary Scott Bessent, have accused Chinese AI companies of using distillation to extract capabilities from US systems. In July, Bessent described the practice as "theft" and threatened sanctions against overseas companies that use it to leverage US-built models.

Regulatory and Industry Accusations

Recent actions by US agencies and companies highlight the severity of the issue:

  • The Cybersecurity and Infrastructure Security Agency (CISA) accused China’s AI companies of conducting "industrial-scale knowledge distillation campaigns" that violate US companies’ terms of use.
  • Anthropic reported earlier this month that it found Alibaba (developer of the Qwen model family) and DeepSeek engaging in "illicit distillation."
  • China has rejected these claims.

Huang suggested that if providers dislike their products being used for training, they should "know your customers, and disable the service," placing the onus on providers to enforce their own terms rather than relying on government intervention.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.