Treasury Secretary Urges Development of Open-Source AI to Counter Chinese Models

Treasury Secretary Scott Bessent testified before the House Financial Services Committee, emphasizing the need for the United States to increase the development of open-source artificial intelligence models. He warned that Chinese firms have been using a technique called "distillation" to copy U.S. closed-source models, creating competitive alternatives at lower cost.

"One of the things I would say is that we need to develop… more open-source models in the U.S. We can't let these large labs have regulatory capture because that will stop innovation," Bessent said.

He added that expanding open-source AI would reduce reliance on Chinese models, which he said sometimes claim to be advanced U.S. systems such as Anthropic’s Mythos or Claude. Bessent noted that Anthropic’s Mythos model, released earlier this year, represents a significant step toward artificial general intelligence, underscoring the strategic importance of home-grown technology.

Industry leaders, including Nvidia, Microsoft, Meta, Dell Technologies, Palantir, Hugging Face, Mozilla and Mistral, have jointly urged the government to avoid broad restrictions on open-weight AI, arguing that targeted legal frameworks can address unlawful copying without stifling innovation. The administration’s voluntary AI framework, issued in August, exempts open-source models from pre-release security reviews, focusing oversight on proprietary systems.