US Bans DeepSeek AI From Civilian Government Systems Over Security Risks

The US Department of Homeland Security has ordered federal civilian agencies to remove all DeepSeek products and third-party services built on its technology from their IT systems. The ban, made public in October but effective since July, prohibits new purchases and mandates the removal of existing implementations across agencies including the Treasury, State, and Justice Departments. This move expands earlier restrictions previously limited to defense, intelligence, and commerce sectors.

Security Concerns

DHS cited "multiple threat vectors," warning that DeepSeek’s models could transfer sensitive data to China due to servers located there and Chinese legal frameworks allowing government access. The order also flagged potential malware and critical vulnerabilities identified in DeepSeek’s iOS application. Agencies may request exceptions within 60 days by submitting written justifications.

Geopolitical Context

The ban intensifies tensions in the US-China AI race. DeepSeek’s low-cost, high-performance models disrupted the industry in early 2025, challenging US dominance. While Washington accuses DeepSeek of illicit model distillation—using American outputs to train its own systems—Beijing opposes what it calls the politicization of trade. A summit between Presidents Trump and Xi is scheduled for November in Shenzhen, where AI cooperation remains a key topic despite ongoing friction.