Discussions have intensified again in Washington about how to limit the spread and use of Chinese open-weight artificial intelligence models. The trigger was the release of the Kimi K3 model by Moonshot AI, which demonstrated impressive results comparable to the best closed-source counterparts.
Pressure Scenarios: From Trade Lists to Regulatory Risks
In 2025, the U.S. Department of Commerce already considered adding several Chinese AI companies to a blacklist, which would require American organizations to obtain special licenses for any interaction with them. Additionally, the National Security Agency (NSA) and the White House Office of Cybersecurity were working on public warnings about potential threats posed by Chinese developers.
Currently, possible tools under discussion include restrictions on government procurement, inclusion in export control lists, and official risk warnings. Earlier, the idea of a presidential executive order imposing responsibility on American companies for security when using such models was abandoned due to fears of slowing down the entire industry's development.
Expert Opinion and White House Reaction
Dean Ball, former senior AI policy advisor in the Trump administration and now head of Strategic Futures at OpenAI, expressed the view that the best strategy for U.S. authorities would be to create a massive regulatory risk for using Chinese open-source models. He emphasized that the quality of Kimi K3 cannot be dismissed as mere knowledge distillation from other systems.
However, White House AI advisor David Sacks categorically rejected this approach, calling the use of regulatory uncertainty as a competitive weapon "absolutely unacceptable." He noted that leading developers of closed systems, such as OpenAI and Anthropic, are effectively forming a duopoly in revenue and may be interested in having the state artificially limit competition from open-source solutions.
China Reciprocates
While U.S. authorities discuss measures, Beijing is also considering imposing its own restrictions on the export of advanced AI technologies. This involves potential control over the transfer of training data and a ban on foreign users downloading model weights. Moreover, Alibaba has already prohibited its employees from using Anthropic products, citing security risks.
My analysis: The situation resembles a classic "security vs. innovation" dilemma. The Trump administration's attempts to block open Chinese models could lead to fragmentation of the global AI landscape, which in the long term would hurt all market participants, including American startups dependent on open-source solutions. The real battle now is not so much over technology as over control of standards and the infrastructure for their distribution.