The US Department of Homeland Security has issued an order mandating the removal of DeepSeek products from federal civilian agency IT systems. Filed on Monday and effective immediately upon its Tuesday publication, the directive extends to DeepSeek affiliates and third-party services built using DeepSeek implementations, while prohibiting affected agencies from making new purchases. DHS cited risks that government use could expose sensitive information regarding US citizens and federal programs to China, alongside identified security vulnerabilities in DeepSeek’s iOS application.
This action expands prior restrictions within the Commerce Department and segments of the defense, security, and intelligence communities amid ongoing competition between Washington and Beijing for artificial intelligence leadership. The move follows similar prohibitions enacted by Australia in February 2025.
The DHS order signals a decisive shift from advisory caution to mandatory operational exclusion for Chinese AI tools within US civilian infrastructure. By explicitly targeting not just direct products but also affiliates and third-party integrations, the department aims to close potential backdoors where DeepSeek capabilities might be embedded indirectly into government workflows. This comprehensive scope suggests regulators view the risk as systemic rather than isolated to specific software versions.
From a Compliance perspective, this creates immediate friction for vendors and contractors who may have integrated DeepSeek-based solutions into their offerings for federal clients. The citation of data exposure risks to China underscores growing geopolitical tensions over AI supply chains, likely prompting further scrutiny of foreign-origin models across allied nations. Agencies must now rapidly audit existing IT inventories to ensure full adherence, potentially disrupting established procurement pipelines.
