China’s Response to the Federal Measure
China has challenged the U.S. ban requiring federal civilian agencies to remove DeepSeek and third-party services built with the company’s technology. The Associated Press reports that the Department of Homeland Security order, made public on Monday, also prohibits new purchases and dates from late July. Beijing said it will defend the rights of Chinese companies and accused Washington of excessively expanding the concept of national security.
The U.S. measure cites the risk that sensitive data could be accessible to Chinese authorities through servers and infrastructure located in China. These are assessments expressed by the U.S. government in the order, not an independent review published by AP of DeepSeek’s handling of every piece of data. The company had not immediately responded to the agency’s request for comment.
The Ban Has a Specific Scope
The measure applies to federal civilian agencies and government IT systems: it is not equivalent to a general ban for all U.S. citizens or companies. The order covers third-party products and services developed using DeepSeek, a broader scope than just the app or website. The distinction matters for organizations that purchase software: an AI component can be embedded in services that do not display the brand of the original model.
DeepSeek has released open-weight models used by developers and organizations in various countries. The availability of the weights makes it possible to run some versions on one’s own infrastructure, but it does not eliminate security requirements or public procurement obligations. Organizations must map dependencies and components, verify where data are processed, and understand which versions are actually present in their systems.
A Rift Within a Broader Relationship
The dispute is part of the competition between Washington and Beijing over models, chips, and access to data. The United States has maintained restrictions on the export to China of certain advanced chips and the equipment to produce them. In parallel, the U.S. administration has accused some Chinese companies of distilling capabilities from U.S. models; Beijing has rejected such accusations. These issues are distinct from the DeepSeek ban, but they contribute to a climate of mutual distrust.
The two powers have also discussed forms of cooperation on AI safety, including a possible notification of incidents related to national security. This makes the situation less straightforward than a simple confrontation between blocs: technical dialogue and restrictive measures can proceed at the same time.
China’s response alone does not change the obligation of U.S. agencies. It does show, however, that public software choices have become instruments of industrial and diplomatic policy. Assessing the concrete risks requires information on data, infrastructure, and controls, beyond the statements of the governments involved.
