US orders federal agencies to remove China’s DeepSeek AI from their systems

US Government Orders Removal of DeepSeek AI

The U.S. government has directed federal civilian agencies to stop using and remove all technology from the Chinese AI startup DeepSeek, citing significant national security risks. According to an order from the Department of Homeland Security, the move is a response to concerns that sensitive U.S. government data could be accessed by Chinese authorities through the company's operations. This directive significantly escalates the ongoing technology competition between the U.S. and China, shifting the focus from hardware to the foundational AI models themselves.

The Technical and Security Concerns

The DHS order is comprehensive, targeting not just DeepSeek's direct products but also any third-party services built upon its AI technology. This extends a previous law from December 2025 that focused on its use in national security systems. The core of the government's concern is data sovereignty and security, amplified by allegations from rivals like OpenAI and Anthropic that DeepSeek has engaged in model distillation—using outputs from American AI systems to train its own. The U.S. formally accused DeepSeek and five other Chinese AI firms of this practice in a September report.

• The directive mandates that federal civilian agencies cease all new purchases of DeepSeek products and services.
• Existing installations of DeepSeek technology must be removed from government IT systems.
• The ban includes third-party applications and services that integrate or use DeepSeek's underlying models.
• Exceptions can be requested by agencies, but are not guaranteed.

Broader Ecosystem Implications

This action against DeepSeek, whose low-cost open-weight models have challenged U.S. competitors, reflects a new front in U.S.-China tech policy. While Beijing has opposed the move as an overreach of national security concepts, it is reportedly also considering its own export restrictions on certain AI technologies. For the global AI market, the U.S. order introduces a new layer of compliance and supply-chain scrutiny. Companies and developers leveraging AI models will now face increased pressure to verify the origin and training practices of their foundational technology to avoid regulatory and operational risks.

The U.S. government's action against DeepSeek moves the front line of the tech rivalry from semiconductor hardware to the AI models themselves, creating significant compliance and supply-chain risks for any enterprise leveraging cross-border AI technology.