The Department of Homeland Security has issued an order banning federal civilian agencies from acquiring products and services developed by Chinese AI company DeepSeek, and requiring them to remove the company's products from government information technology systems. The move, reported Monday by Reuters, CyberScoop and other outlets, marks the most sweeping federal action against the Chinese AI lab to date.
What the Order Requires For more context on this story, see our ongoing AI news.
According to PYMNTS, which reviewed the notice, the order was posted on the Federal Register's public inspection site on October 19 and is scheduled for publication in the Federal Register on October 20. It invokes the Federal Supply Chain Security Act of 2018, which allows DHS to manage supply chain risks by recommending that risky products or services be excluded from federal procurement or removed from federal IT systems.
The order directs civilian agencies on two fronts: they may not procure DeepSeek products or services, and they must strip existing DeepSeek software from their IT environments. Reporting from Startup Fortune indicates the restrictions extend to government contracts as well, cutting off indirect routes through which the company's technology could reach federal systems.
The Threat Vectors DHS Cited
DHS described DeepSeek as a privately owned company located in the People's Republic of China whose models "present multiple threat vectors." The notice pointed to three specific concerns: Chinese laws that allow the government to access private company infrastructure, an account-generation process that collects personally identifiable information, and a prompt-and-response design that requires users to submit their information to the model in order to use it.
"The transfer of the aforementioned information to the PRC, either independently or in combination with publicly available data, could produce actionable intelligence on U.S. persons and/or government programs," DHS said in the order, according to PYMNTS.
The department went beyond data flows, raising the possibility of compromised software itself. "The models themselves may contain malware or other security vulnerabilities," the order stated. "Indeed, open-source security research has found that the DeepSeek iOS mobile application contains multiple critical vulnerabilities."
Why Now
DeepSeek's own trajectory explains why regulators keep returning to the company. Reuters reported as far back as February 2025, when the company's chatbot surged to prominence and became one of the most downloaded AI applications worldwide, that its rise was met with growing security concerns prompting multiple governments to ban or restrict it. Since then the company has kept shipping: its newest model, the V4.1-Flash-1017 update released earlier this week, arrived with open weights under an MIT license and strengthened agentic coding performance, extending the low-cost strategy that made DeepSeek a household name in the first place. Engadget noted that the federal ban lands just days after that release.
The federal order also follows a long pattern of government friction with DeepSeek. Several US states, along with agencies such as NASA and the US Navy, had already prohibited the company's tools on government devices. What distinguishes the new order is its breadth: rather than device-level bans, it uses supply chain security authority to bar the technology from civilian agency IT wholesale.
A Broader US-China AI Rift
The ban arrives amid deteriorating rhetoric between Washington and Beijing over AI. US authorities have accused Chinese AI companies, including DeepSeek and Moonshot AI, of running industrial-scale campaigns to distill American frontier models, allegations the companies have disputed. China, for its part, has probed its own companies over data flows to Western AI providers and tightened export controls on AI technologies.
For DeepSeek, the direct commercial impact of the order may be limited, since it does not sell meaningfully to the US government. The larger consequence is symbolic and regulatory: the US government now formally classifies one of the world's most prominent open-weight AI providers as a supply chain risk, a designation that enterprise customers and allied governments tend to watch closely.
What Enterprises Should Take Away
Security teams outside government will read the order for its risk framing. DHS's argument rests not on evidence that DeepSeek models have been weaponized, but on structural risk: jurisdiction, data collection practices, and documented vulnerabilities in the mobile app. That framework parallels guidance other agencies have issued about evaluating AI vendors on data handling and software assurance rather than model quality alone.
The order takes effect through standard Federal Register publication, and agencies will now begin compliance work under the 2018 statute's exclusion and removal processes. That mechanism is deliberately procedural: DHS reviews products and services for supply chain risk, issues an exclusion or removal recommendation, and agencies are expected to act on it — a process designed to move faster than legislation and to apply across the civilian enterprise without requiring each agency to run its own security review.
For DeepSeek, the practical effects are threefold. Federal sites must remove the company's apps and integrations from managed systems, procurement teams must treat DeepSeek offerings as excluded from new contracts, and vendors bundling third-party AI into government-facing products will need to show their supply chains are clean. Compliance burdens of exactly this kind tend to ripple outward, as contractors align their own security postures with what federal buyers are permitted to run.
Whether further action follows against other Chinese AI providers remains an open question, but the message to federal buyers is unambiguous: Chinese AI models are now treated as a procurement risk at the highest level of the US government.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
