The U.S. government has expanded restrictions on Chinese AI startup DeepSeek by ordering all federal civilian agencies to stop using its products and services, citing national security risks linked to data handling and potential Chinese government access under local laws.
- Order applies to DeepSeek products and third-party services using its technology
- Concerns focus on user data accessibility by Chinese government under local laws
- China condemns US actions as politicizing technology and trade issues
What happened
The U.S. Department of Homeland Security has directed all federal civilian agencies to cease purchasing and to remove existing DeepSeek AI products and services from their systems. This includes technologies and services built on DeepSeek’s platform, although agencies may seek exceptions under certain conditions. The directive addresses mounting concerns over national security, particularly regarding DeepSeek’s data handling practices. DHS warns that user data from government systems potentially could be accessed by Chinese authorities under China’s legal framework.
This order expands on earlier restrictions that were imposed in December 2025 limiting DeepSeek’s use in intelligence and national security agencies. The broader civilian agency ban now covers both purchases and deployment of the Chinese AI firm’s offerings, signaling escalating caution toward foreign AI technologies amid rivalry with China. DeepSeek gained industry attention with its advanced R1 reasoning model launched in early 2025, which has since sparked intense scrutiny from U.S. regulators and competitors alike.
Why it matters
DeepSeek represents a critical node in the growing global AI competition between the United States and China. Its open-weight models challenge established U.S. AI players by providing competitive capabilities at lower costs, heightening concerns about sensitive data security and intellectual property protection. The U.S. government’s action reflects fears that using technology from Chinese origins could expose classified or sensitive information, given China’s laws that might compel companies to cooperate with state authorities.
Multiple Western AI developers, including OpenAI and Anthropic, have raised allegations of DeepSeek illegally extracting data from their models through a method called model distillation, where outputs from one AI system are used to train another. Such practices exacerbate the perceived risks tied to DeepSeek’s involvement in American AI ecosystems. The U.S. move also aligns with China’s parallel steps to restrict AI-related technology exports amid its own security concerns, indicating a deepening decoupling in AI innovation and governance.
What to watch next
Attention will focus on how federal agencies comply with and enforce the new removal mandate, especially regarding existing contracts and embedded DeepSeek technologies. The availability of exceptions could become a point of negotiation or legal challenge. Furthermore, the impact of this move on broader U.S.-China tech relations is likely to influence future AI regulatory frameworks and international cooperation efforts.
On the global stage, China’s response defending its enterprises and criticizing the U.S. for politicizing security issues may prompt reciprocal measures or stalled collaboration. Meanwhile, ongoing investigations into model distillation accusations and intellectual property concerns could spur additional regulatory scrutiny for Chinese AI firms operating internationally. Market watchers should also monitor emerging Chinese limits on AI technology exports as these could reshape supply chains and innovation dynamics.