As reported by BBC News on October 19, 2026, the United States Department of Homeland Security (DHS) has ordered all artificial intelligence technologies developed by Chinese developer DeepSeek removed from civilian government systems. The binding federal directive instructs executive departments to dismantle active deployments of DeepSeek models, while explicitly barring the use of software tools or algorithmic pipelines built with the assistance of the Beijing lab's models.
The directive applies across civilian cabinet agencies, including the Department of the Treasury, the Department of State, and the Department of Justice, though agencies operating under the Department of Defense are exempted from the immediate mandate. While the DHS policy was enacted internally in late July 2026, federal officials confirmed the directive publicly on Monday as congressional committees intensify oversight into artificial intelligence supply chains and national security risks.
The regulatory move reflects escalating geopolitical tension surrounding open-source AI capabilities. DeepSeek disrupted the global artificial intelligence landscape by open-sourcing frontier-class reasoning and coding models developed at an estimated training cost of $6 million—a fraction of the $100 million or more invested in comparable commercial systems by Silicon Valley laboratories. The resulting cost efficiencies prompted aggressive enterprise adoption, with US cloud distributors including Snowflake, Vercel, and OpenRouter hosting DeepSeek models for commercial software developers.
Federal scrutiny has been fueled by intellectual property concerns voiced by domestic AI developers. Both OpenAI and Anthropic have raised allegations that rival international laboratories systematically harvest output traces from their proprietary models to execute high-throughput synthetic distillation. In an internal White House policy memorandum circulated earlier this year, technology lead Michael Kratsios warned of malicious foreign exploitation targeting American AI development, arguing that unvetted foreign foundation models introduce severe supply-chain vulnerabilities into federal workflows.
While federal civilian procurement of DeepSeek is now barred, the directive does not constrain private enterprise deployment within the United States. However, legal scholars observe that the ban marks a precedent for categorical government blacklists of foreign foundation models, signaling that future regulatory frameworks may mandate rigorous provenance audits for open-weight software across critical infrastructure sectors.