US Orders DeepSeek AI Removed From Most Civilian Government Systems
Homeland Security bars use of the Chinese model and tools built with it, excluding Defense Department agencies, as US-China tensions over AI continue
By AXY Tech Desk ·
Key points
- The US Department of Homeland Security has banned all use of DeepSeek AI within civilian government agencies, including any technology built with its models.
- The order was made public on Monday but took effect from the end of July, and it excludes Department of Defense agencies.
- The DHS has not detailed the specific security concerns behind the ban.
- The ban follows criticism of Chinese AI by the Trump administration and American firms, including copying claims made by Anthropic and OpenAI.
- DeepSeek, the DHS and the Chinese embassy in Washington were contacted for comment.
The United States has ordered that artificial intelligence (AI) technology from the Chinese firm DeepSeek be removed from most government systems, according to a Department of Homeland Security (DHS) order made public on Monday. The DHS has banned all use of DeepSeek within civilian government agencies, along with any technology built with the aid of its AI models. The ban took effect from the end of July, the BBC reported in a story published on 19 October. The decision deepens a technology rivalry between Washington and Beijing at a time when both governments regard AI as central to economic and military strength.
The order covers civilian agencies, including the Department of the Treasury, the Department of State and the Department of Justice. Agencies within the Department of Defense are excluded. The DHS notice did not detail the exact security concerns behind the decision. The DHS, DeepSeek and the Chinese embassy in Washington DC have been contacted for comment.
What DeepSeek is and why it has drawn attention
DeepSeek is a large language model, the type of AI system that generates text, answers questions and writes software code in response to a user's instructions. Models of this kind are trained on very large volumes of data and require substantial computing power to run. The BBC describes DeepSeek as a powerful, popular and relatively inexpensive AI model from China. It is also open source, a term that in this case means the model is free to access and less expensive to use than many competing systems.
That combination has made DeepSeek a disruptive presence in the global race to build advanced AI. When a capable model is cheap and freely available, developers, businesses and researchers can adopt it quickly, including people far outside the country where it was built. The model has grown in popularity since its initial public release in 2025, and demand has been strong enough that the company's servers have gone down at times.
Popularity is central to why governments pay attention. A widely used AI system can receive large amounts of input from its users, and it can be built into other software products. Security officials generally examine such tools in terms of what data they might expose, who controls the underlying systems and whether the software could behave in ways its users cannot easily detect or control. The DHS has not publicly said which of these risks applies in this case, so the reasons behind the order remain unconfirmed.
How the ban works in practice
For civilian federal agencies, the practical effect is that staff are barred from using DeepSeek's models and from using technology built with their help. Because the order extends to tools built with the models, its reach could go beyond the chatbot itself. Agencies would need to identify where such tools appear in their systems, which can be difficult when modern software is assembled from many components, some of them supplied by outside vendors.
The exclusion of the Department of Defense is significant. It suggests the order is directed at civilian government networks rather than military systems, which operate under their own rules and procurement processes. The source material does not explain why the Defense Department was excluded, and no reason has been given publicly in the material available for this report.
The order also has implications beyond federal offices. Government agencies are large buyers of technology in most countries, and a federal restriction can influence how private companies assess whether a product is acceptable to use. The source material does not say whether the order applies to private individuals or businesses, and it does not describe any penalties for non-compliance, so the scope of its enforcement remains unclear.
The political backdrop in Washington
The ban follows criticism of Chinese AI from the Trump administration and from American technology companies. The White House has criticised AI firms from China this year, according to the BBC. US president Donald Trump has made clear that he sees China as the key American rival in the development of AI, which he regards as crucial to the country's future.
The administration's concerns have also appeared in writing. White House tech lead Michael Kratsios wrote in an internal memo earlier this year that "malicious exploitation" of American AI firms had been occurring, mainly by rival firms in China. The wording points to a claim that some Chinese companies are benefiting from work done by US developers rather than building their systems independently. The memo was described in the BBC report, and the full text was not published in the material available.
The move also fits a longer pattern of US restrictions on Chinese technology in government and sensitive settings, justified on national security grounds. Such restrictions typically aim to reduce the risk that software or hardware controlled from abroad could be used to collect information from government networks. The DeepSeek order follows that approach, although it targets one AI company and its models rather than a broad category of products.
Claims of copying by American AI companies
The concerns extend to the private sector. Anthropic and OpenAI, both of which are poised to list on the public stock market in the coming months, have claimed that their AI models were essentially copied by some Chinese AI firms, including DeepSeek. According to the BBC, these claims come from companies preparing for public listings, a period in which they face heightened scrutiny of the risks to their businesses.
The kind of copying alleged is often described in the industry as distillation. In general terms, a developer trains a new model on the outputs of an existing one, which can reproduce much of the original system's ability at a fraction of the cost of building it from scratch. Whether that happened in any specific case is a separate question, and the BBC report does not include evidence for the particular claims about DeepSeek. The claims are attributed to the companies that made them, and the material available to AXY News does not include a response from DeepSeek to them.
If such practices were proven, they could raise questions about intellectual property, meaning the legal protection for inventions, software and other creative or technical work. They could also prompt debate about how AI models should be shared or accessed across national borders. Those questions are unresolved, and no court or regulator has settled them in the material reviewed for this report.
Earlier US steps against Chinese technology
The DeepSeek order is not the first time Washington has limited the use of Chinese technology in government. In 2018, a US defence law barred federal agencies from buying telecommunications equipment from Huawei and ZTE, two Chinese companies, citing national security concerns. In 2023, federal agencies were directed to remove the video app TikTok, which is owned by the Chinese company ByteDance, from government-issued devices. These measures were built on similar arguments about data access and foreign control.
What distinguishes the DeepSeek order is the subject. AI models are not simply hardware or applications that store data. They process what users type into them and can shape the answers and code they produce, which makes the questions about trust and control more complex. Analysts who follow technology policy have argued that this is one reason AI has become a focus of national security debates. The source material does not describe any direct link between the DeepSeek order and the earlier measures, so this comparison is analysis rather than a reported connection.
The order is also consistent with the wider rivalry described in the BBC report. Washington and Beijing are both investing heavily in AI, and each sees leadership in the field as an economic and strategic advantage. Restrictions on rival technologies are one way governments seek to protect their own industries and sensitive systems, but they can also raise costs for companies and users who rely on global supply chains.
What to watch next
Several questions remain open. The DHS has not yet detailed the security concerns behind the order, and the material available does not say whether further guidance will follow. DeepSeek, the DHS and the Chinese embassy in Washington have been contacted for comment, and any replies could clarify the company's position and the government's reasoning. The material available does not include a response from the embassy.
Analysts would be likely to watch whether other federal departments issue similar instructions, and whether Congress or the White House sets out a broader framework for Chinese AI tools. Any move to extend restrictions to private companies or to the Defense Department would represent a significant escalation and would likely draw attention from technology firms, trade officials and allies.
The next phase may also depend on the planned public listings of Anthropic and OpenAI. Those filings and investor briefings could bring more detailed disclosures about the copying claims and the security risks the companies face. For now, the DHS order stands as a clear signal that Washington treats Chinese AI models as a national security question for government use, even as the underlying technology continues to spread quickly around the world. Readers should watch for official clarification from the DHS, any statement from DeepSeek, and any change in how federal agencies apply the rule in the weeks ahead.
Questions and answers
Why did the US ban DeepSeek AI from government use?
The Department of Homeland Security has not publicly detailed the security concerns behind the ban. The order follows criticism of Chinese AI from the Trump administration and American firms, including claims that some Chinese companies copied American AI models.
Which US government agencies are affected by the DeepSeek ban?
The ban applies to civilian agencies, including the Department of the Treasury, the Department of State and the Department of Justice. Agencies within the Department of Defense are excluded.
When did the DeepSeek ban take effect?
The DHS made the order public on Monday, but the ban took effect from the end of July. The source material does not specify the year the ban began.
What is DeepSeek?
DeepSeek is a powerful, popular and relatively inexpensive open-source artificial intelligence model from China. It is free to access and less expensive to use, and it grew in popularity after its initial public release in 2025.
Topics: DeepSeek, Department of Homeland Security, United States, China, Artificial intelligence, Michael Kratsios, Anthropic, OpenAI
Sources: BBC Technology