The US Department of Homeland Security (DHS) has issued a directive banning all civilian federal agencies from using AI products developed by the Chinese company DeepSeek. Agencies are required to remove any existing software associated with the firm from their systems.
The ban applies to both direct DeepSeek products and third-party systems built using its technologies. The directive's language remains ambiguous regarding whether it covers software that only accesses DeepSeek models via API or code generated by the company's models.
This measure follows earlier restrictions implemented by other US government entities. In January 2025, the Defense Information Systems Agency (DISA), the US Navy, NASA, and the House of Representatives began blocking DeepSeek tools within their IT infrastructures.
Concerns regarding DeepSeek have persisted since the release of its DeepSeek-R1 model in early 2025. While the company claims the model cost $294 thousand to train, US lawmakers have labeled the technology a national security threat, citing potential data collection risks.
