Homeland Security Orders DeepSeek Products Removed From Federal Civilian Systems

The Department of Homeland Security has ordered federal civilian agencies to stop buying products from the Chinese artificial intelligence company DeepSeek and to remove those already in use from government computer systems. The order became public on Monday and was published in the Federal Register on Tuesday, according to TechCrunch and BleepingComputer.

The order covers DeepSeek’s own products, those of its affiliates, and third-party products and services built on DeepSeek technology. Agencies must take the products off approved lists and purchasing marketplaces and remove them from their systems immediately, or ask for an exception within 60 days, with a justification and a plan to reduce the risk. The Cybersecurity and Infrastructure Security Agency will review those requests and can provide technical help to agencies that lack the capacity to comply, BleepingComputer reported.

The order applies only to civilian agencies under the department’s authority. It is not a general ban on private or consumer use of DeepSeek’s tools.

In explaining the decision, the department said sensitive data from government users could become accessible to Chinese authorities under Chinese law. It also raised concerns about possible malware in DeepSeek models and cited a February 2025 analysis by the security firm NowSecure of DeepSeek’s iPhone app, which reported unencrypted data transmissions, weak encryption and insecure storage of user data.

China objected. Foreign Ministry spokeswoman Mao Ning said Beijing opposes overstretching the concept of national security and politicizing technology and trade issues, and pledged to protect the rights of Chinese companies, TechCrunch reported. That argument carries some weight: the order rests on what Chinese law could allow and on security weaknesses researchers have described, not on a public finding that government data has actually been taken.

DeepSeek drew global attention in January 2025, when the release of its R1 reasoning model drew wide industry notice and set off a sharp sell-off in Nvidia’s shares. Governments and regulators have scrutinized the company since. A law passed in December 2025 restricted its use in intelligence agencies’ national security systems, and the new order extends limits to civilian agencies.

The order also lands amid a wider dispute over how Chinese labs build their models. A government advisory in September accused DeepSeek and several other Chinese AI companies of large-scale distillation, a technique that uses another system’s answers to train a new model. OpenAI told members of Congress earlier this year that it had seen suspected distillation attempts by DeepSeek, and Anthropic has said it detected unauthorized efforts by Chinese labs to collect outputs from its Claude models, TechCrunch reported.

For agencies, the practical challenge may lie in finding DeepSeek technology inside products sold under other names, which is why the order’s reach to third-party services is likely to matter most.