US Bans DeepSeek From Federal Civilian Systems Over Security Concerns

The United States has ordered federal civilian agencies to remove Chinese artificial intelligence company DeepSeek’s products from their information technology systems, citing concerns that the tools could expose sensitive government data and create cybersecurity risks.

The US Department of Homeland Security (DHS) also prohibited civilian government agencies from purchasing DeepSeek products and third-party services built using the company’s AI models, according to a notice published on October 19. The order covers products developed by DeepSeek and its affiliates.

The move extends Washington’s efforts to restrict the use of Chinese technology across government systems, as concerns over data security and national security increasingly influence decisions about which AI tools public institutions can use.

Why the US Is Targeting DeepSeek

The DHS described DeepSeek’s models as presenting “multiple threat vectors”, warning that their use by government employees or contractors could result in sensitive information being transferred to China.

The department also raised concerns that the technology could generate actionable intelligence about US individuals or government agencies. It warned that the models themselves could contain malware or other security vulnerabilities.

These are risks cited by the US government as the basis for its decision. The published notice does not, by itself, establish that DeepSeek has transferred sensitive US government information to Chinese authorities or that its models contain malware.

In a memo issued in July and made public on October 19, the DHS had also instructed federal agencies to monitor traffic moving into and out of government websites to address potential risks associated with DeepSeek.

The Ban Extends Beyond DeepSeek's Own Products

The restrictions are not limited to applications developed directly by DeepSeek.

Under the order, agencies must remove third-party products and services developed using DeepSeek implementations. They are also prohibited from acquiring covered products and services for civilian government use. The measure was issued under the Federal Acquisition Supply Chain Security Act of 2018, which provides a legal framework for addressing risks in government technology supply chains.

The scope matters because organisations can incorporate AI models into other software and services without necessarily offering the underlying model as a standalone product.

By including third-party services built using DeepSeek, the order addresses a wider range of potential government technology deployments.

The restrictions apply to federal civilian government systems and procurement. They should not be interpreted as a blanket ban on DeepSeek for every US business or individual.

DeepSeek Has Faced Earlier Restrictions

The latest order follows previous restrictions on DeepSeek within parts of the US government.

The Department of Commerce and the US Navy had already prohibited the use of DeepSeek on government devices. Several states, including New York, Tennessee and Virginia, had also introduced similar restrictions.

The measures form part of a broader US effort to limit potential security risks associated with Chinese technology.

The Federal Communications Commission is also scheduled to vote on October 29 on a proposal to prevent Chinese laboratories from testing electronic devices intended for the US market. The proposed restrictions would affect products such as smartphones, cameras and computers.

Together, these actions illustrate how national security concerns are increasingly shaping US technology procurement and regulation.

What the Decision Means for AI Competition

DeepSeek's rise has intensified competition in the global AI industry, challenging established assumptions about the resources required to develop capable AI models.

Its technology has attracted international attention by offering an alternative to models developed by major US AI companies. But the growing availability of AI tools has also created questions for governments about data handling, software security and the risks of deploying foreign-developed systems in sensitive environments.

For public institutions, the decision to use an AI model involves more than its performance or cost. Where sensitive government information is involved, agencies must also consider where data may be processed, what information a system can access and whether its software can be independently assessed.

The US restriction makes that distinction clear: a model's technical capabilities do not automatically make it suitable for government use.

However, the order does not establish that every DeepSeek deployment is unsafe. It reflects the US government's assessment of the risks associated with using the company's products within federal civilian systems.

A Wider Challenge for Governments

The DeepSeek restrictions highlight a dilemma facing governments as AI becomes more widely integrated into public services and administrative systems.

AI tools can help organisations analyse information, automate routine tasks and improve productivity. Yet deploying them in government environments can introduce additional concerns when systems handle confidential records, sensitive communications or operational data.

Governments must therefore weigh the benefits of adopting AI against the risks associated with data protection, software integrity and dependence on external technology providers.

For the US, the latest order places DeepSeek outside the permitted technology supply chain for federal civilian agencies. It also reinforces a broader shift towards treating AI security as part of national security policy.

As competition between the United States and China continues, decisions about which AI systems governments can use are likely to remain closely tied to questions of trust, data control and technological dependence.