Washington has barred Chinese AI technology from civilian government agencies, citing wider concerns over security and competition as the US seeks to maintain its lead in artificial intelligence, reports Asianlite News Desk.
The United States has ordered the removal of artificial intelligence (AI) technology developed by Chinese firm DeepSeek from most government systems, escalating tensions in the global race for AI dominance.
The Department of Homeland Security (DHS) has prohibited the use of DeepSeek across civilian government agencies, including the Departments of the Treasury, State and Justice. The restriction also covers technology developed using DeepSeek’s AI models.
However, agencies within the Department of Defense are excluded from the order.
The DHS made the directive public on Monday, although the ban took effect at the end of July. The department has not detailed the specific security concerns behind the decision. The DHS, DeepSeek and the Chinese embassy in Washington DC have been approached for comment.
The move comes amid growing criticism of Chinese AI companies by the Trump administration and US technology firms. Earlier this year, White House technology lead Michael Kratsios wrote in an internal memo that American AI models were being subjected to “malicious exploitation”, primarily by rival companies in China.
US firms Anthropic and OpenAI have also accused Chinese AI developers, including DeepSeek, of effectively copying their models.
President Donald Trump has repeatedly identified China as Washington’s principal competitor in AI development, which his administration views as strategically important to America’s economic and technological future.
Despite the concerns, DeepSeek has become a significant player in the AI industry since releasing its models publicly in 2025. Its free chatbot application has ranked among the most downloaded apps on Apple’s App Store, while heavy demand has occasionally disrupted its servers.
DeepSeek’s open-source models are also used by American technology businesses, including Snowflake, Vercel and OpenRouter. Their accessibility and relatively low operating costs have made them attractive to startups and developers.
Unlike proprietary systems such as Google’s Gemini, Anthropic’s Claude and OpenAI’s Codex, DeepSeek’s models can be accessed more cheaply, reducing the cost of building AI-powered products.
The company has also challenged assumptions about the expense of developing advanced AI. DeepSeek previously said training its debut model cost about $6m (£4.5m), compared with an estimated $100m for OpenAI’s GPT-4. The disclosure triggered a sharp sell-off in technology stocks, the BBC reported.
DeepSeek released its latest model last week, drawing praise from some users who said it performed competitively against advanced AI tools on tasks such as software coding.
The ban underscores the growing tension between national security concerns and the commercial appeal of cheaper Chinese AI technology.
