US bans DeepSeek AI in federal civilian agencies

Federal civilian agencies must stop buying and remove DeepSeek AI from government systems, citing risks tied to user data and operations in China.

US bans DeepSeek AI in federal civilian agencies

In October 2026, the U.S. Department of Homeland Security ordered federal civilian agencies to stop purchasing DeepSeek products and services and remove the Chinese AI startup’s technology from government IT systems. TechCrunch reported the order on October 20, citing DHS, and said it covers both future purchases and products already used in agency systems.

Key facts

  • The restriction applies to federal civilian agencies and covers both new purchases and DeepSeek technology already present in government IT systems.
  • Third-party products and services built using DeepSeek technology are also covered.
  • Agencies can request exceptions to the order, according to DHS.
  • DHS cited concerns about DeepSeek’s handling of user data, warning that sensitive information from U.S. government users could potentially be accessible to Chinese authorities under Chinese law.
  • A law passed in December 2025 had already targeted DeepSeek’s use in national-security systems, according to TechCrunch.

Why it matters

The broader scope is the main operational change for agency IT teams. Because the order reaches third-party products and services built using DeepSeek technology, removing a direct DeepSeek account or app would not necessarily address every covered use. Agencies must also identify services that incorporate the startup’s models without presenting themselves as DeepSeek products.

That turns DeepSeek into both a product question and a software-supply-chain question: agencies have to review direct tools, embedded AI features and the vendors responsible for them.

DHS’s stated concern centers on data handling. The order says sensitive information from U.S. government users could potentially be accessible to Chinese authorities under Chinese law. For private users and companies, the sources describe a federal civilian-agency restriction rather than a nationwide prohibition on using DeepSeek.

Order details

TechCrunch’s summary of the DHS directive sets out four main elements:

Order elementReported effect on agencies
New purchasesStop buying services and products developed by DeepSeek.
Existing systemsRemove DeepSeek technology from government IT systems.
Third-party productsCover products and services built using DeepSeek technology.
ExceptionsAllow agencies to request an exception to the requirements.

The wording makes vendor documentation and system inventories important for implementation because the restriction reaches more than a standalone chatbot. It also means an agency cannot assume that the absence of DeepSeek branding on an application automatically places that service outside the order.

Background

DeepSeek drew broad attention in January 2025 when its R1 reasoning model entered the center of the AI industry, according to TechCrunch. The report described the Chinese AI lab as challenging U.S. rivals including OpenAI and Meta with low-cost, open-weight models. In general, an open-weight model makes its trained parameters available for others to download or build on rather than limiting access to a hosted interface.

Model distillation is a separate technical issue mentioned in the reporting. The technique uses outputs from one AI model to train another. According to TechCrunch, a September U.S. report accused DeepSeek and five other Chinese AI companies, including Alibaba and Moonshot AI, of using large-scale distillation to extract information from American AI systems.

OpenAI and Anthropic have also described their own findings. OpenAI told U.S. lawmakers it had observed suspected attempts by DeepSeek to distill its models, while Anthropic said it had identified unauthorized attempts by DeepSeek and other Chinese AI labs to collect outputs from Claude. These are allegations and company assessments rather than a finding described by TechCrunch.

That distinction is important: the DHS directive focuses on data handling and DeepSeek’s operations in China, while the distillation claims concern how AI models may be developed.

What’s next

Implementation is the immediate issue. Agencies will need to identify direct and embedded DeepSeek use, remove covered products from government systems and decide whether to seek exceptions. The exception process is especially relevant for third-party services because the order applies even when DeepSeek is not the visible brand.

Further updates should focus on removal progress, exception requests and any technical guidance defining what counts as a covered third-party product. The sources did not provide a removal timetable or full exception criteria.

FAQ

Does this order prevent everyone in the United States from using DeepSeek?

No. It directs federal civilian agencies. TechCrunch’s report did not describe a prohibition for private users or companies.

Does it cover third-party apps that use DeepSeek technology?

Yes, according to the DHS summary. The order covers third-party products and services built using DeepSeek technology, although agencies can request exceptions.

Why did DHS cite Chinese law?

DHS warned that sensitive information from U.S. government users could potentially be accessible to Chinese authorities under Chinese law. The separate claims from OpenAI and Anthropic concern alleged model-distillation activity rather than the data-access warning in the order.

Sources: Tech Xplore — electronic gadgets, technology advances and research news, TechCrunch.