The Trump administration is currently grappling with a significant divide regarding its strategy for handling the rapid advancements in Chinese artificial intelligence (AI). According to multiple sources familiar with ongoing discussions, officials within the White House advocate for stricter controls on Chinese AI systems that threaten to compete with leading U.S. models. Conversely, the Commerce Department views such measures as impractical.
This internal debate has escalated following the announcement from China’s Moonshot AI lab about its new Kimi K3 model, which reportedly rivals top models from U.S.-based companies like Anthropic and OpenAI. Discussions among senior White House officials have turned to potential actions aimed at hindering the development of Chinese AI by limiting their ability to train on U.S. models—a process known as distillation.
The urgency for a response has intensified after Anthropic accused the Chinese tech giant Alibaba of executing a “distillation attack” against its Claude Fable 5 model, leading to heightened scrutiny from the White House. While officials are bracing for some form of presidential action in light of these developments, it is not anticipated to materialize as an executive order.
Recognizing the seriousness of distillation attacks, the White House has been closely monitoring developments. A recent announcement revealed that Moonshot AI’s Kimi K3 was developed using Anthropic’s technology, prompting Treasury Secretary Scott Bessent to describe such covert distillation practices as akin to intellectual property theft, suggesting the possibility of sanctions.
Despite the ongoing discussions, formal requests for input or guidance have yet to be communicated to the Commerce Department, which has developed its own perspectives on the issue. The Commerce Department plays a key role in export controls and has gained influence in this policy debate, pushing for measures that encourage the development of comparable open-weight models by U.S. labs as a counter to China.
As the administration navigates this complex landscape, the overarching challenge lies in striking a balance between regulatory measures and fostering innovation within domestic AI labs. In recent weeks, attention has shifted to Anthropic’s Claude Mythos and Fable 5 models due to their capabilities in identifying vulnerabilities in government systems, further complicating the administration’s AI policy landscape amid the rise of unsecured open-weight models from China.
