U.S. Bars DeepSeek AI from Federal Civilian Systems Amid Escalating China Tech Tensions

The U.S. government has mandated the removal of all artificial intelligence technology from Chinese firm DeepSeek across most federal systems.

The Department of Homeland Security (DHS) has prohibited the use of DeepSeek within civilian government agencies, including any technology developed using its AI models.

As an open-source AI model—freely accessible and comparatively inexpensive—DeepSeek has emerged as a disruptive force in the global AI race.

Although the DHS notice did not specify the exact security concerns tied to DeepSeek, the White House has repeatedly criticized Chinese AI firms this year.

The DHS, DeepSeek, and the Chinese embassy in Washington, D.C., have been contacted for comment.

The DHS announced the order on Monday, though the ban took effect at the end of July.

The directive applies specifically to civilian agencies, such as the Department of the Treasury, the Department of State, and the Department of Justice. Agencies under the Department of Defense are excluded.

The ban follows heightened criticism of Chinese AI from the Trump administration and American industry leaders.

White House technology advisor Michael Kratsios wrote in an internal memo earlier this year that “malicious exploitation” of U.S. AI firms was occurring, primarily by rival companies in China.

Anthropic and OpenAI, both preparing for public stock offerings in the coming months, have also alleged that certain Chinese AI firms, including DeepSeek, have essentially copied their models.

President Donald Trump has repeatedly identified China as the United States’ primary rival in AI development, which he views as critical to the nation’s future.

DeepSeek has surged in popularity since its initial public release in 2025, with demand so high that its servers have occasionally crashed.

While its free chatbot app ranks among the most downloaded on the Apple App Store, DeepSeek’s models are also integrated into services by U.S. tech firms such as Snowflake, Vercel, and OpenRouter.

Startups have increasingly turned to open-source models like DeepSeek.

For developers, building with DeepSeek costs a fraction of what closed alternatives—such as Google’s Gemini, Anthropic’s Claude, or OpenAI’s Codex—would require.

DeepSeek has stated that it spent approximately $6 million (£4.5 million) to train its debut model, compared to roughly $100 million for OpenAI’s GPT-4. The disclosure contributed to a selloff in technology stocks.

The latest version of DeepSeek was released last week. Users have praised it online, claiming it outperforms even OpenAI’s most advanced tools on common tasks such as software coding.

Source link