The US DHS orders all federal civilian agencies to remove DeepSeek's tech from their IT systems and prohibits future procurement of its products and services

Federal civilian agencies must now remove and cease purchasing DeepSeek AI products, increasing immediate IT management overhead.

The U.S. Department of Homeland Security (DHS) has ordered all federal civilian agencies to remove DeepSeek's artificial intelligence technology from their IT systems and prohibits any future procurement of its products and services. The directive, set to be published in the Federal Register, cites national security concerns, including the potential for data transfer to servers in China, risks of malware, and overall threats to federal IT infrastructure. Secretary of Homeland Security Markwayne Mullin stated that DeepSeek models pose "multiple threat vectors," with specific concerns about data collection and privacy due to Chinese national security laws that grant the government broad access to data held by domestic companies. The ban comes after a review by the Federal Acquisition Security Council, and agencies are encouraged to enhance their security controls post-removal.

This directive signals a heightened governmental scrutiny of AI technologies originating from geopolitical rivals, particularly China. The DHS's action establishes a precedent for how national security concerns, data sovereignty, and potential state-sponsored espionage via AI will shape federal procurement and technology adoption moving forward. It also highlights the vulnerability of open-weight models and the challenge of ensuring data privacy when AI systems are hosted in jurisdictions with differing legal frameworks regarding government access to data. Agencies will need to invest in robust vetting processes for AI tools to mitigate these risks.

The ban directly impacts agencies that may have integrated DeepSeek's capabilities into their operations, necessitating immediate remediation efforts and potentially disrupting workflows. For other AI vendors, especially those with ties to China or operating under similar data handling models, this serves as a clear warning and may lead to increased compliance burdens and a review of their own data practices to avoid similar restrictions. The move also underscores a growing trend towards the "securitization" of AI, where national security considerations are becoming as important as performance and cost in technology evaluations.