Sino-US AI model distillation technology sparks controversy: Bessant calls it theft, Huang calls it competition

The AI ​​war between China and the United States is igniting new gunpowder! Bessant denounced "distillation" as theft, and Huang Renxun said it was competition. (File photo)

[Financial Channel/Comprehensive Report] Artificial intelligence (AI) model "distillation" technology has triggered a new round of controversy between China and the United States. U.S. Treasury Secretary Scott Bessent has described the use of distillation technology to extract capabilities from U.S. AI models as "theft" and warned that sanctions may be imposed on relevant overseas companies. However, NVIDIA CEO Jensen Huang has a different view and believes that training or learning from competitors' products is "competition."

In an interview with CNBC's "Squawk Box" on Monday, Huang was asked whether AI model distillation could be considered "robbery," and said, "That's called competition." He pointed out that companies can test other companies' products and even "reduce the product to its skeleton" to understand how it works.

Please read on...

The so-called AI model distillation is to use the output results generated by other AI models to train your own model, so that the new model can learn the capabilities of the existing model. This technology has now become a sensitive issue in the AI ​​competition between China and the United States. U.S. officials have accused some Chinese AI companies of using distillation methods to obtain relevant capabilities from U.S. AI systems.

Huang Renxun also admitted that he certainly hopes that other companies will not learn from Huida products, "so that we can have a smooth journey." But he also said competition makes everything better. He also pointed out that if a company does not want others to use its products, the most direct way is to understand the customers and shut down related services.

The White House did not immediately respond to CNBC's request for comment.

The U.S. Cybersecurity and Infrastructure Security Agency earlier this month also accused Chinese AI companies of conducting "industrial-scale knowledge extraction activities" and said that related actions violated the terms of use of U.S. companies.

In addition, AI company Anthropic said earlier that Alibaba, which developed the Qwen series of models, and DeepSeek were involved in what it called "illegal distillation" activities. China denies the accusations.

Grasp the economic pulse with one hand. Click here to subscribe to Free Finance Youtube channel