US Orders Agencies to Pull DeepSeek AI Tools from Government Devices

By Elena Vasquez

The US government on Monday ordered its agencies to stop using AI tools from Chinese startup DeepSeek, with the Department of Homeland Security warning the tools "may result in the transfer of sensitive data to (China)". The directive came from the Department of Homeland Security in a notice that also covers tools created by third parties using DeepSeek technology, and it requires their removal from government IT devices.

This is not a soft advisory. It is a removal order. Agencies must strip DeepSeek products, and anything built on top of them, from the federal IT estate — a ban that reaches beyond the startup's own apps to cover derivatives built by other companies.

The move lands as the US and China race for dominance in AI, a sector both governments treat as a strategic and economic priority. Washington's decision to blacklist a foreign model from federal systems signals how AI tools are now judged through the same national security lens as telecommunications equipment and semiconductors.

What exactly did the Department of Homeland Security order?

The DHS notice, issued Monday, announces a ban on two categories of products:

• Products developed by DeepSeek itself
• Tools created using DeepSeek by third parties

Both categories must be removed from IT devices used by US government agencies. That second category is the part with real bite. It means a vendor cannot launder DeepSeek's underlying technology into a rebranded product and sell it into federal agencies. The ban follows the model, not the label on the box.

The Department of Defense weighed in on the rationale. Use of the tools, the defense department said, "may result in the transfer of sensitive data to (China)". The quote, as reported, keeps the phrasing verbatim — and the parenthetical "(China)" is doing deliberate work, naming the state that US agencies fear would sit at the end of that data pipeline.

Why does the DeepSeek ban matter beyond government IT?

The context here is the race between the US and China for dominance in what the source calls the fast-moving AI sector. DeepSeek is a Chinese startup that has become one of the most watched names in that race. When a government bans a foreign AI tool from every agency's devices, it is making a statement about where it believes data flows and who might eventually see it.

The stakes run in two directions.

First, there is the data question. Government agencies handle sensitive information by default. If an AI tool can move that data offshore — even indirectly, through a third-party product built on DeepSeek — the DHS has now decided that risk is unacceptable inside federal IT.

Second, there is the market question. A federal ban sets a precedent. Third-party developers now know that building on DeepSeek locks them out of US government contracts entirely. That calculus could reshape which foundation models vendors choose to build on, not just in Washington but across the wider US software market that sells into government.

What does the ruling change?

Practically, it changes three things at once.

• Removal, not restriction. Agencies must take DeepSeek-based tools off IT devices. There is no carve-out for approved use cases mentioned in the directive as reported.
• Liability extends to derivatives. Third-party tools built using DeepSeek fall under the same ban. Vendors, not just DeepSeek itself, carry the compliance burden.
• The security framing is now explicit. The defense department's statement puts a named risk — sensitive data transfer to China — at the center of the policy.

The ban also fits a broader pattern in how Washington handles Chinese technology. Where earlier waves of restriction targeted hardware and networking gear, this order targets AI models directly — software that can be embedded invisibly inside other products. That is precisely why the third-party clause exists. Regulators cannot police what they cannot see, so the rule reaches down the supply chain to catch DeepSeek wherever it hides.

Who is DeepSeek, and why this company?

DeepSeek is a Chinese startup working in artificial intelligence, and it has become a flashpoint in the US-China competition for dominance in the sector. That competition, as the source notes, is fast-moving, and Monday's order shows the US government treating Chinese AI models as a live national security concern rather than a commercial curiosity.

The defense department's warning is the sharpest public articulation of that concern. By saying use of the tools "may result in the transfer of sensitive data to (China)", the department framed the risk not as hypothetical spying but as a data governance failure waiting to happen inside federal systems.

For DeepSeek itself, the US government market is now closed. The bigger consequence is reputational and commercial: a formal DHS ban, backed by a defense department data warning, gives every US enterprise procurement team a ready-made reason to exclude DeepSeek-based tooling from vendor reviews.

What comes next?

The order takes its place inside the wider US-China race for AI dominance, and it will not be the last word. Expect other governments, and US state agencies, to study the DHS notice and the defense department's data-transfer rationale as they weigh their own rules for foreign AI tools. For vendors, the message from Monday is blunt: build on DeepSeek, and the US federal market is off the table.

---

The ban was announced in a Department of Homeland Security notice on Monday, covering DeepSeek products and third-party tools built using DeepSeek, which must be removed from government IT devices.