US Department of Homeland Security Orders Ban on DeepSeek in Government Systems

US orders ban on DeepSeek across government systems: US-China AI competition continues to heat up

The US has ordered the removal of all artificial intelligence (AI) technology from Chinese company DeepSeek from most government systems.

The US Department of Homeland Security (DHS) has banned civilian government agencies from using DeepSeek, as well as any technology developed using its AI models.

Amid the militarization of US-China AI, BBC learns the Pentagon has formally stopped using Anthropic tools, October 6, 2026

Trump rules out US-China cooperation on AI development, September 30, 2026

Anthropic co-founder calls for mandatory AI "emergency shutdown" mechanism, September 15, 2026

End of content

BBC has contacted the Department of Homeland Security, DeepSeek and the Chinese Embassy in the United States to request comment.

The Department of Homeland Security announced the order on Monday (October 19), but the ban has been in effect since late July.

The order applies to civilian agencies, including the Treasury Department, the State Department and the Justice Department, but not to agencies under the Department of Defense.

Riots in France: Partial curfew in Paris, looted goods appear on second-hand platforms

Riots across France: 19th-century historic building burned down, politicians' statements reflect divisions

Zunjie V800 "brake gate": From a product safety dispute to a "hating the country" controversy

After Hong Kong actors Chung Suet-ying and Yuen Lai-lam spoke out, they recounted their experiences of gender-based offense

End of Trending

White House technology chief Michael Kratsios wrote in an internal memo earlier this year that US AI companies have been "maliciously exploited," mainly by competitor companies in China.

Anthropic and OpenAI are both expected to go public in the coming months. Both companies also claim that some Chinese AI companies, including DeepSeek, have effectively copied their AI models.

US President Donald Trump has made clear that in AI development he sees China as the United States' main competitor, and he believes this is crucial to America's future.

Since its first public launch in 2025, DeepSeek's popularity has continued to rise, with strong demand for its models even causing the company's servers to crash at one point.

DeepSeek's free chatbot app has topped the Apple App Store download rankings, and its AI models have also been adopted by US tech companies. Companies including Snowflake, Vercel and OpenRouter have integrated DeepSeek into the services they offer customers.

Startups are also adopting open-source models like DeepSeek.

For tech companies or AI developers, using DeepSeek at work costs only a fraction of closed-source models such as Google Gemini, Anthropic Claude or OpenAI Codex.

Similarly, the cost DeepSeek bears in developing its models is far lower than the money US AI companies invest in developing their own models.

Last year, DeepSeek said training the first version of its model cost about $6 million, while the related cost of OpenAI GPT-4 was about $100 million. These disclosures sent tech stocks falling sharply.

DeepSeek's latest version was released last week. Users of the model have praised it online, saying it even outperforms OpenAI's most advanced AI tool on common AI tasks such as computer software programming.

This article was originally written in English. We used artificial intelligence to assist with translation, and a BBC journalist reviewed it before publication. Learn more about how we use artificial intelligence.