Both.news
Donate
Just the Facts

US orders civilian agencies to remove DeepSeek AI products from systems

The Homeland Security order bars DeepSeek products and tools built on its models from federal civilian systems and contracts, citing security risks the department says include sensitive data reaching China.

Both News · AI-written from cited reporting

Written from 7 reports by outlets across the spectrum

Updated:

The Department of Homeland Security has ordered U.S. civilian government agencies to remove products from the Chinese artificial intelligence startup DeepSeek from their systems and to stop buying them, a directive made public on Monday.3,4,5,6

The order also covers tools that third parties have built using DeepSeek's models, and it bars government contracting for those products. In a document attached to the notice, the department said that "DeepSeek's models present multiple threat vectors." It warned that use of DeepSeek AI tools "may result in the transfer of sensitive data to [China] if used by US government employees or vendors."3,4,5

The document said DeepSeek's servers are located in China, where, it said, the legal system permits the government "extensive authority" to access private company infrastructure. The department's main worry is that sensitive government data could end up in China, giving Chinese intelligence services a window into American citizens and government programs.3,5

The directive applies to civilian agencies. The department can issue direction to civilian agencies but not to the military or intelligence services, so the Department of Defense is excluded. Agencies including the Treasury, State and Justice departments are covered. The Defense Department did not respond to a request for comment from AFP.3,5,6

The timing of the order is described differently across coverage. The Hill said the order initially went out in July, and Breitbart said it was issued on July 24 but made public only this week. The BBC said the ban took effect from the end of July, while the Guardian described the government as having ordered its agencies to stop using the tools on Monday.3,4,5,6

Agencies that need more time, or want an exception, can submit a written justification to the department within 60 days of the order, including requests for compliance delays. The BBC said the notice did not detail the exact security concerns surrounding DeepSeek. Breitbart reported that the department pointed to open-source research that found several critical vulnerabilities in DeepSeek's iOS app.5,6

DeepSeek, one of China's top AI labs, shook up the industry in January 2025 with a low-cost, high-performance chatbot that challenged assumptions of U.S. dominance in the field. The company released a long-awaited new model, V4, in April, and earlier in October it updated a faster, more lightweight version of V4. The BBC said the most recent version was released last week.3,6

The BBC reported that DeepSeek said it spent about $6 million to train the debut version of its model last year, while the costs associated with OpenAI's GPT-4 were roughly $100 million. Its free chatbot app is among the most downloaded in the Apple App Store rankings, and U.S. companies including Snowflake, Vercel and OpenRouter incorporate DeepSeek in their offerings to customers.6

The U.S. has accused DeepSeek and other top Chinese AI firms of systematically stealing the capabilities of cutting-edge models developed by American rivals. White House tech lead Michael Kratsios wrote earlier this year in an internal memo that "malicious exploitation" of American AI firms had been occurring, mainly by rival firms in China. Anthropic and OpenAI have also claimed that their AI models were essentially copied by some Chinese AI firms, including DeepSeek. President Donald Trump has said that he sees China as the key U.S. rival in the development of AI.3,6

China's foreign ministry responded to the ban. Spokesperson Mao Ning, responding to a Reuters inquiry, said China "consistently opposes the over-generalization of national security as well as the politicization of science, technology and trade issues." She added that Beijing would safeguard the rights and interests of its companies.5

DeepSeek was contacted for comment, and the BBC said it had also contacted the Department of Homeland Security and the Chinese embassy in Washington.3,6

The ban comes as the two governments continue to discuss AI while they compete on it. When Xi Jinping and Donald Trump met at the White House in September, China and the U.S. agreed to set up a "communication channel" for AI incidents. The topic is expected to arise again at a meeting between Xi and Trump at a summit scheduled for November in the Chinese city of Shenzhen.3

How each side told it

Left

The Guardian led with the Monday notice and the countries' race for dominance in the sector, and its page also carried an extended passage on Trump administration actions toward the press and on news ownership that the other reports did not include.

Center

CNBC's headline placed the ban within an "AI rift with China," The Hill emphasised the contracting bar and the order's July origin, and the BBC put the action in the context of the AI race, the Kratsios memo and DeepSeek's popularity among users and U.S. companies.

Right

Breitbart led with the administration's ban and the department's concern about espionage, cited the open-source vulnerability research and an earlier interview in which an author advocated blocking the app, and included China's foreign ministry response, while the New York Post headline used the framing "cracks down" and "intensifies" and the Washington Times headline framed the action as Homeland Security banning DeepSeek from federal civilian systems.

Sources

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7

Written by AI from the reports listed below and nothing else. Loaded words are taken out, every claim is attributed to whoever made it, and each fact links to the outlets that report it. If something looks wrong, check the sources and tell us.