The US government bans DeepSeek: Chinese AI accused of mass surveillance

On Monday, October 19, a measure from U.S. Secretary of Homeland Security Markwayne Mullin was published. It prohibits the country’s civilian agencies from buying and using DeepSeek products.A measure that directly targets the Chinese state, accused of several misdeeds through this artificial intelligence.

The United States no longer wants DeepSeek on its territory

It was the Department of Homeland Security (DHS) that took this measure under the Federal Acquisition Supply Chain Security Act of 2018. This law allows suppliers deemed risky to be excluded from the federal supply chain, which also covers computer applications and systems. The order was therefore signed by the department’s secretary, Markwayne Mullin: it was initially dated July 24, before being publicly published on October 19 in the Federal Register, according to the BBC. This comes one step after the U.S. Defense Information Systems Agency (DISA) blocked DeepSeek on Pentagon networks in January 2025, as noted by CyberScoop. In May 2025, it was Microsoft that prohibited its employees from using this AI.

This measure imposes two obligations on civilian agencies: end all current uses of DeepSeek and refrain from installing it again. It directly covers the version of DeepSeek distributed by Hangzhou DeepSeek Artrificial Intelligence Co., the Chinese company that develops it, but does not stop there. It also concerns all third-party applications based on DeepSeek. This includes those using the open-source version of the chatbot, which anyone can download, modify, and use.Only agencies under the Department of Defense can continue using this AI: they fall under a different framework in the government’s view.

Why Washington is targeting this Chinese AI

The DHS justifies this exclusion by citing several cybersecurity and data sovereignty risks. It explains that because DeepSeek’s servers are in China, the government could mass-collect U.S. government data. This is the core of the DHS’s argument: the Chinese government could compel DeepSeek to give it access to its servers and monitor users, thereby generating actionable intelligence on U.S. citizens and government programs.

According to the agency, DeepSeek’s models present “several threat vectors”: possible collection of personal data, user tracking, security vulnerabilities, and even the potential presence of malware. Another argument in favor of the ban is that, with its R1 model, the chatbot had shaken up Washington and Silicon Valley, since the model performed similarly to the best American models (GPT, Gemini, etc.) while claiming a much lower development cost. DeepSeek was subsequently accused, notably by the White House, of using American LLMs to train itself.A hypothetical argument, since it has not been claimed by the Department of Homeland Security.

This is not the first time a Chinese digital service has been banned internally by the U.S. government. A law adopted in 2022 and which entered into force in 2023 had banned TikTok from federal phones. A ban that was lifted this year: the Department of Justice determined that the American version of TikTok was no longer covered by the law, since it operates independently of ByteDance, its Chinese publisher.

👉🏻 Follow the latest tech news in real time: add 01net to your sources on Google, and subscribe to our WhatsApp channel.

Source: Le Parisien