The US Department of Homeland Security has ordered federal civilian agencies to remove all technology made by the Chinese artificial intelligence company DeepSeek from their systems, widening Washington's push against Chinese AI tools.
The order was made public on Monday, but BBC News reported that the restriction had taken effect at the end of July. It covers DeepSeek's own products and any technology built with the help of its models.
It applies to civilian departments including the Treasury, the State Department and the Department of Justice. Agencies inside the Department of Defense are excluded.
The ban extends beyond DeepSeek's chatbot. Civilian agencies must also stop using third-party products and services that were built with DeepSeek technology, which could affect tools supplied by outside vendors.
The notice does not set out every security concern in detail. The Verge reported that officials are worried the models may transfer sensitive data to China. DeepSeek keeps data on servers in China, and US officials have argued that Chinese law gives Beijing broad powers to demand information from domestic companies.
DeepSeek became one of the most talked-about names in AI after releasing powerful open-source models that cost far less to run than closed systems from OpenAI, Anthropic and Google. Its free chatbot has ranked among the most downloaded apps, and several US technology companies have incorporated its models into their products.
The move follows months of criticism of Chinese AI from the Trump administration and from American AI firms. OpenAI and Anthropic have both accused Chinese companies, including DeepSeek, of copying their models. BBC News said DeepSeek, the Department of Homeland Security and the Chinese embassy in Washington had been asked for comment.


