China Calls U.S. Ban on DeepSeek an 'Overstretching' of National Security

By Nathan Brooks

The U.S. Department of Homeland Security has ordered all federal civilian agencies to purge DeepSeek's products from their IT systems — and Beijing is firing back. China's Foreign Ministry on Tuesday called the move an "overstretching of the concept of national security."

The DHS notice, made public Monday in the Federal Register, instructs agencies to remove DeepSeek and its affiliated products, plus third-party products and services built using DeepSeek technology. The order, which dates back to late July, also blocks civilian agencies from procuring DeepSeek products and services going forward.

For context on scale: DeepSeek is one of China's best-known AI companies, and its open-source models are widely available across many countries — including the United States — and mostly free to use. That's precisely what makes this ban different from typical procurement blacklists: the underlying models sit inside countless third-party tools, so the order reaches far beyond one vendor's contracts.

What does the ban actually allege?

The Federal Register notice claims DeepSeek's models "present multiple threat vectors." The stated concerns are concrete: DeepSeek's servers are located in China, and the notice flags the risk that Chinese authorities could potentially access private company infrastructure and sensitive U.S. data through that arrangement.

DeepSeek did not immediately respond to a request for comment.

In Beijing, Foreign Ministry spokesperson Mao Ning was blunt when reporters asked about the ban at a regular press conference. China, she said, "will resolutely safeguard the legitimate rights and interests of Chinese companies."

"China consistently opposes the overstretching of the concept of national security and the politicizing of science, technology, economic and trade issues," Mao told reporters at the routine briefing.

What does this change for practitioners?

If you build on open-source models, read the third-party clause carefully. The order doesn't just cover DeepSeek's own apps — it sweeps in any product or service developed using DeepSeek. Federal contractors and vendors serving civilian agencies will need to audit their model supply chains to prove compliance, and the compliance burden lands on them, not on DHS.

For everyone else, the signal is clear: Washington is treating Chinese-hosted model infrastructure as a data-exposure risk by default, regardless of whether a model is free or open-source.

The backdrop: a fragile AI truce

The dispute lands weeks before U.S. President Donald Trump and Chinese President Xi Jinping meet in mid-November on the sidelines of the APEC leaders' summit in Shenzhen, China. AI is expected to be on the agenda.

The two countries have recently made progress on some AI governance collaboration, including a U.S. proposal for a notification system on national security-related incidents. But they remain at odds on many issues.

The friction points are stacking up:

• U.S.-led restrictions still bar China from accessing some of the world's most advanced AI chips and the machines that make them.
• Last week, Beijing rolled out new proposed plans to restrict exports of certain advanced AI models.
• The U.S. has accused Chinese AI startups, including DeepSeek, of large-scale "distillation" — extraction of capabilities from frontier American AI models — claims Beijing has dismissed.
• Earlier this year, the U.S. Federal Communications Commission barred imports of new foreign-made humanoid robots, a move widely seen as targeting China.

The DeepSeek ban shows the limits of that cooperation: governance talks can proceed while procurement walls go up. Expect Shenzhen to test whether both can continue at once.