The United States has ordered artificial intelligence technology from Chinese firm DeepSeek to be removed from most government systems. The Department of Homeland Security (DHS) issued the order.
The ban covers all use of DeepSeek within civilian government agencies. It also covers any technology built with the help of DeepSeek’s AI models.
The DHS made the order public on Monday. However, the ban has been in effect since the end of July.
Which Agencies Are Affected
The order applies to civilian agencies, including the Department of the Treasury, the Department of State and the Department of Justice. Agencies within the Department of Defense are not included.
The DHS notice did not explain the exact security concerns about DeepSeek. The DHS, DeepSeek and the Chinese embassy in Washington DC have been contacted for comment.
The ban follows criticism of Chinese AI from the Trump administration and from American companies. The White House has spoken out against Chinese AI firms this year.
White House tech lead Michael Kratsios wrote in an internal memo earlier this year that “malicious exploitation” of American AI firms had been taking place. He said this was mainly by rival firms in China.
Anthropic and OpenAI have both claimed that some Chinese AI firms, including DeepSeek, essentially copied their AI models. Both companies are expected to list on the public stock market in the coming months.
President Donald Trump has said he sees China as the key US rival in AI development. He views AI as crucial to the country’s future.
Why DeepSeek Has Grown So Popular
DeepSeek has grown in popularity since its first public release in 2025. It is an open source AI model, which means it is free to access and cheaper to use than many rivals.
Its free chatbot app is one of the most downloaded apps in the Apple App Store rankings. Demand for its models has at times been so high that the company’s servers have gone down.
US tech companies also use DeepSeek. Snowflake, Vercel and OpenRouter all include DeepSeek in products they offer to customers.
Startups have been drawn to open source models like DeepSeek. For developers, using DeepSeek costs a fraction of what closed models such as Google’s Gemini, Anthropic’s Claude or OpenAI’s Codex would cost.
DeepSeek has also spent far less to build its models than US AI firms. Last year, the company said it cost about $6 million (£4.5 million) to train the first version of its model.
By comparison, the costs linked to OpenAI’s GPT-4 were roughly $100 million. When DeepSeek’s figures came out, tech stocks fell sharply.
DeepSeek released the latest version of its model last week. Users have praised it online, saying it beats even the most advanced OpenAI tools on common tasks such as computer coding.
