BEIJING (Realist English). On October 20, Chinese Foreign Ministry spokesperson Mao Ning stated at a regular briefing that China consistently opposes the overgeneralization of the concept of national security and the politicization of trade, economic, and technological issues, and will resolutely protect the legitimate rights and interests of Chinese companies.
This statement was a response to the US Department of Homeland Security’s demand that federal civilian agencies stop using products from the Chinese artificial intelligence company DeepSeek.
Content of the US ban
On October 19, the US Department of Homeland Security published a notice requiring federal civilian agencies to remove from their information systems products developed by DeepSeek, as well as third-party services built on DeepSeek models, and prohibiting the procurement of relevant products.
The notice claims that DeepSeek’s models pose “multiple threats”: if US government employees or contractors use its AI tools, this “could lead to the transfer of sensitive data to China.” It also notes that DeepSeek’s servers are located in China, and local legislation grants the state “broad powers” to access the infrastructure of private companies.
The ban applies to civilian agencies, including the Treasury Department, the State Department, and the Justice Department, but does not affect structures subordinate to the Pentagon. According to the BBC, the order was made public on October 19 but has been in effect since late July under the Federal Acquisition Supply Chain Security Act of 2018.
The rise of DeepSeek and the controversy around it
DeepSeek is one of China’s leading AI laboratories. In January 2025, it caused a sensation in the industry by introducing a chatbot with low cost and high performance, challenging the perception of US dominance in AI.
Its open models can be deployed for free, and the cost of use is significantly lower than that of closed models from American companies OpenAI and Anthropic. According to the BBC, training the first version of the DeepSeek model cost approximately $6 million, while the cost of OpenAI GPT-4 is estimated at approximately $100 million.
However, the US has accused DeepSeek and other leading Chinese AI companies of “systematically stealing” advanced model capabilities from American competitors. Earlier, the head of technology at the White House stated in an internal memo that American AI companies are being subjected to “malicious exploitation,” primarily by Chinese competitors.
This ban is another escalation of US-China competition in AI. Unlike the sanctions against Huawei in the field of chips, the threat from DeepSeek is defined as a “data flow”: the AI model itself may become a channel for collecting and transmitting information.
If this logic gains broad acceptance, it will affect all AI services with servers in China, not just DeepSeek. For Chinese companies, this means further raising regulatory barriers to operating in the US, and the “free” advantage of open models may be offset by geopolitical costs.
For the global AI ecosystem, this accelerates the trend toward the formation of technological barriers between the US and China: third-party developers, when choosing a base model, will be forced to consider geopolitical affiliation, rather than only comparing performance and cost.
