US threatens sanctions on Chinese AI models over IP theft

On Tuesday, Treasury Secretary Scott Bessent signaled that the U.S. would scrutinize open-source AI models from China for intellectual property theft, threatening sanctions if violations are found. Speaking on Fox Business, Bessent stated that while the administration supports open-source models, it will not tolerate IP theft. The statement comes as Chinese models like Moonshot AI's Kimi K3 gain ground, potentially harming U.S. AI giants like OpenAI and Anthropic. Axios also reported that the Trump administration is considering a wholesale ban on Chinese open-source models, though this has been disputed. This represents an escalation in the U.S.-China AI race, following export controls on advanced chips.

Model distillation—a technique transferring capabilities from large to smaller models—is at the heart of the debate. Microsoft CEO Satya Nadella criticized large U.S. labs for imposing restrictive terms on distillation after benefiting from fair-use training on public data. Meanwhile, Anthropic recently settled a $1.5 billion lawsuit over unauthorized use of copyrighted books. However, Hugging Face CEO Clem Delangue argued that distillation is only a minor factor, emphasizing China's strong research teams and collaborative approach as key to its AI advances, not just IP theft.