US orders federal agencies to remove DeepSeek AI tools

The U.S. Department of Homeland Security has ordered all federal civilian agencies to immediately remove DeepSeek AI products and services from their systems, according to an order signed by Homeland Security Secretary Markwayne Mullin on July 24, 2026, and publicly filed on October 19, 2026. The directive applies to products made by Hangzhou DeepSeek Artificial Intelligence and any third-party tools developed using DeepSeek's AI models, but excludes military and intelligence agencies.

The order cites multiple security concerns, including that DeepSeek's servers are based in China where local law permits government access to private company infrastructure. The DHS summary warns that DeepSeek's sign-up process collects personal information that could be used to track and digitally fingerprint users, and that sensitive data entered by U.S. government staff or contractors could reach Chinese servers and be combined with public data to produce actionable intelligence. The order also references a February 2025 security audit that identified critical flaws in DeepSeek's iPhone application. The order's evidence is drawn from DeepSeek's own privacy policy, a 2020 DHS business advisory on Chinese data risks, and an April 2025 House Select Committee on China report.

Agencies have a 60-day window to request exceptions from the Cybersecurity and Infrastructure Security Agency (CISA) with compelling justifications such as mission-critical impact. Until full compliance, agencies must monitor network traffic, increase logging, implement data controls, and flag unusual activity. DeepSeek, released with open-weight models that can be freely deployed at low cost, has gained significant traction since its public launch in early 2025, with its latest R1 model widely praised for performance on coding and other AI tasks. The order will be reviewed annually.

China's foreign ministry spokesperson Mao Ning condemned the ban on October 20, stating that China "consistently opposes overstretching the concept of national security and politicizing science, technology, economic and trade issues." Mao pledged that China would "resolutely safeguard the legitimate rights and interests of Chinese companies." The ban reflects broader U.S.-China competition in AI, with Chinese competitors like DeepSeek offering lower-cost alternatives to American companies such as OpenAI and Anthropic, which rely primarily on proprietary or closed models.