US Department of Homeland Security Orders Removal of DeepSeek AI Technology from Federal Civilian Agencies

Key summary

The US Department of Homeland Security (DHS) has issued an order to remove artificial intelligence technology from the company DeepSeek from all federal civilian agencies. Published on the Federal Register, the directive warns that DeepSeek's AI models present 'multiple threat vectors' and could grant the Communist Party of China 'extensive access' to US government infrastructure. Although DeepSeek did not rank in the top 10 of a recent public opinion survey of AI models, its free application saw millions of downloads by Americans in 2025, raising significant concerns for the US government. At that time, a complete ban on the app was under consideration. A cybersecurity expert stated that DeepSeek's privacy policy is 'not worth the paper it was written on.' The new DHS order mandates that information technology systems within all civilian agencies must remove products related to DeepSeek, including third-party services developed using the app. Purchasing products and services from DeepSeek in these agencies is prohibited effective immediately. DHS justified the action by citing the Federal Acquisition Supply Chain Security Act of 2018 following a review by the Federal Acquisition Security Council. The agency warned that the technology poses various dangers to the US, potentially being used to 'track or create digital fingerprints' of users and transfer sensitive data to the People's Republic of China if used by the US government or vendors. The removal effort began last year after DeepSeek surprised the tech industry with its R1 model launch. Despite training on less efficient chips at a fraction of the cost of some competitors, this open-source system outperformed models developed by American rivals OpenAI and Anthropic in several complex tests. The Department of Commerce banned DeepSeek from government devices last year, and the US Navy ordered all personnel to stop using the model due to security concerns. New York State also banned DeepSeek from government devices and networks due to 'serious concerns' regarding data privacy risks and potential censorship. Recently, DeepSeek launched a new AI model named V4.1 Flash-1017, which promises improved speed along with enhanced reasoning and coding capabilities. Preliminary reports suggest it can compete with some of the most advanced US AI systems.