DHS bans DeepSeek AI from federal agencies

On Monday, the U.S. Department of Homeland Security issued a notice that all federal civilian agencies must remove DeepSeek products from their IT systems and are barred from buying new ones. The directive followed the release of DeepSeek’s newest model, V4.1‑Flash‑1017, which the startup called smarter, faster, and more efficient. The order also prohibits the company from participating in any federal government contracts.

Why DHS raised the alarm

DHS said the models could contain malware and other security vulnerabilities. It warned that DeepSeek’s technology might transfer sensitive data to the People’s Republic of China, creating multiple threat vectors. The agency cited the model’s storage of data on servers in China, where national security laws allow the government to request user information, and its collection of personally identifiable data and digital fingerprints.

Agencies are required to delete any existing DeepSeek software and cannot procure new products. They can apply for exceptions, but must submit a request at least 60 days in advance and provide a compelling justification, such as a mission‑critical need. DHS encourages agencies to strengthen security controls, monitor traffic to and from DeepSeek services, and enhance logging and data‑exfiltration detection.

The ban is part of a broader federal push against Chinese technology. The Commerce Department, the Navy, and several state governments have already restricted DeepSeek use. The Federal Communications Commission will vote on October 29 to bar Chinese labs from testing electronics for U.S. use, adding to the pressure on the AI model.

Hangzhou‑based DeepSeek launched its open‑source model in January 2025, offering a low‑cost alternative that rivals high‑end U.S. systems. Its V4‑Pro and V4.1‑Flash releases come at a fraction of the price of competitors, and the company has faced allegations of sending U.S. user data to China and using restricted Nvidia chips in its training.