US Federal DeepSeek AI Ban: What Agencies Need to Know

What are the Security Risks of DeepSeek-R1?

Department of Homeland Security has officially barred civilian federal agencies from purchasing or using products and services from the Chinese AI developer Hangzhou DeepSeek Artificial Intelligence Co. and its affiliates. This directive, which requires agencies to remove DeepSeek software from their IT systems, was signed on July 24, 2026, but only became public on October 19, 2026.

The order was filed for public inspection on October 19 and was scheduled for publication in the Federal Register on October 20, 2026. This action follows a July 2026 memo from the department that had already instructed agencies to monitor traffic flowing to and from government websites for potential risks associated with the software. While DeepSeek-R1, released in January 2025, became a widely downloaded freeware app in the U.S., its use is now prohibited within the federal civilian infrastructure. Italy’s data protection authority previously ordered DeepSeek to block its chatbot in early 2025, citing insufficient responses to privacy concerns.

Is DeepSeek banned in the US for private citizens?

The Department of Homeland Security states that DeepSeek models present multiple threat vectors that jeopardize national security. These models may result in the transfer of sensitive data to the People's Republic of China, as the company operates under laws that allow the government to access private infrastructure. The DHS further warns that the collection of personally identifiable information during account creation and the potential for malware or security vulnerabilities pose significant risks.

By using these tools, agencies could inadvertently provide the Chinese government with actionable intelligence on U.S. persons and government programs. Neither DeepSeek nor the Chinese Embassy in Washington provided immediate comment regarding these security concerns.

How does the DHS DeepSeek directive affect federal IT?

The ban is enforced through Exclusion and Removal Order #2026-01, which DHS Secretary Markwayne Mullin signed on July 24, 2026, under the Federal Acquisition Supply Chain Security Act of 2018. This order applies to all civilian federal IT systems and covers third-party products or services built on DeepSeek implementations.

National security and classified information systems are excluded from this specific order. These systems remain subject to separate directives issued by the Department of War and the Director of National Intelligence. The removal requirements take effect immediately for all other applicable agencies unless they secure an official exception.

Why was the DeepSeek AI ban implemented?

Agencies must remove all DeepSeek software immediately. If an agency requires continued use, it must submit a written request for an exception within 60 days, providing a compelling justification and a plan for proposed mitigations. Until the software is fully removed, agencies are directed to monitor network traffic, strengthen logging and data exfiltration controls, and flag any anomalous activity.

What previous restrictions were already in place?

Several government entities restricted DeepSeek before this order. The Navy, the Commerce Department, and the House of Representatives had already implemented device-level bans on the software. NASA cut off access to the platform in January 2025.

The Pentagon’s Defense Information Systems Agency also blocked the software on military networks in January 2025, a decision made after discovering that Department of Defense employees had used the app on government devices for approximately two days. State-level restrictions were also in place, with New York, Tennessee, and Virginia limiting use by their personnel.

These earlier actions followed a market reaction on January 27, 2025, when DeepSeek caused Nvidia’s stock to drop 16%, erasing approximately $600 billion of the chipmaker’s value and contributing to an estimated $1 trillion loss across U.S. tech. This current order does not add DeepSeek to the Commerce Department’s Entity List.

How Does the DeepSeek AI Model Function?

DeepSeek is a series of large language models developed by a Chinese company of the same name, with the series beginning in November 2023. The model is open-source and can be run locally on your own hardware or hosted by various providers.

For users of the Deepseek.com services, the developers host a version of the model on servers located in China.

What are the concerns regarding DeepSeek's data practices?

Concerns regarding the software stem from allegations made by a House select committee on the Chinese Communist Party in April 2025, which claimed that DeepSeek secretly transmits American user data to the Chinese government. That same committee alleged that DeepSeek may have utilized approximately 60,000 Nvidia chips, including 20,000 units that were restricted under U.S. export controls.

The Department of Homeland Security has cited open-source research that identified multiple critical security vulnerabilities within DeepSeek’s iOS mobile application.

What is the next step for US-China tech regulation?

The agency is scheduled to vote on October 29, 2026, regarding whether to bar Chinese laboratories from testing electronics intended for U.S. use.