The Department of Homeland Security has ordered federal civilian agencies to remove DeepSeek technology from their IT systems and stop purchasing the Chinese artificial intelligence company’s products and services, citing risks to sensitive government information. The restrictions also cover third-party products and services developed using DeepSeek technology.
The order was issued on July 24, 2026, and made public in October. It applies to civilian agencies, including the Treasury, State and Justice departments, but does not cover the Department of Defense. It is a government technology restriction, not a nationwide prohibition on consumers or private businesses using DeepSeek.
The directive’s reach extends beyond the company’s own chatbot and other directly branded offerings. Products from affiliates and outside vendors can also fall within its scope if they use covered DeepSeek implementations. That means agencies assessing compliance must consider the technology underlying software they buy, rather than relying solely on a supplier’s name or a product’s branding.
DHS identified several potential security risks, including the possibility that Chinese authorities could gain access to sensitive U.S. government data. The department cited DeepSeek’s operations and servers in China, alongside concerns about the collection of personally identifiable information, user tracking and the potential transfer of government information to China.
The department also pointed to cybersecurity research identifying critical vulnerabilities in DeepSeek’s iOS application. Its assessment warned that DeepSeek technology could expose agencies to malware or other security weaknesses. Those warnings describe potential threats; they do not establish that sensitive federal data has actually been transferred to China or that the company has committed the misconduct described as a risk.
The order draws on the Federal Acquisition Supply Chain Security Act of 2018. Agencies were given 60 days from its issuance to submit written requests for exceptions, including requests for additional time to comply. That window was tied to the July order, rather than its public disclosure in October.
China rejected the measure. Foreign Ministry spokesperson Mao Ning said Beijing opposed expanding national security arguments into technology and trade restrictions and would defend Chinese companies’ legitimate interests. DeepSeek did not immediately respond to CNBC’s request for comment.
The directive expands a series of U.S. restrictions on DeepSeek in government. Legislation passed in December 2025 targeted its use in intelligence community national security systems. The latest action reaches federal civilian agencies more broadly and addresses both existing systems and future purchases.
DeepSeek attracted international attention in January 2025 with its R1 reasoning model and has competed with U.S. developers through low-cost, open-weight models. Open-weight distribution allows others to obtain model parameters and incorporate the technology into their own offerings, making the origins of an AI component relevant even when an agency purchases software from a different vendor.
For federal suppliers, the directive therefore raises a concrete compliance question: whether their products incorporate covered DeepSeek technology. A product does not have to carry the DeepSeek name to be subject to removal or procurement restrictions.
Sources: ExecutiveGov, Gadget Review, CNBC, TechCrunch