The United States has directed federal civilian agencies to halt purchases of services and products from Chinese AI startup DeepSeek and to remove them from government IT systems over national security concerns linked to the company's operations in China, according to the Department of Homeland Security. The directive extends beyond banning DeepSeek's own offerings to cover third-party products and services built using DeepSeek technology, though agencies may request exceptions.

The DHS highlighted risks around DeepSeek's data handling, noting that sensitive information from US government users could become accessible to Chinese authorities under local laws, the order stated.

The action occurs amid rising AI technology competition between the US and China. The Chinese AI model competes with US rivals such as OpenAI and Meta through low-cost open-weight models.

China responded critically to the US action. Foreign Ministry spokesperson Mao Ning stated Tuesday that Beijing opposes stretching the idea of national security and turning technology and trade matters political, according to an official briefing. The spokesperson added that Beijing will firmly protect the legitimate rights and interests of Chinese enterprises.

The US had already moved to restrict DeepSeek in government settings. A law enacted in December 2025 addressed its use in intelligence and national security systems. The new order broadens these limits to federal civilian agencies and applies to both fresh purchases and existing IT systems.

At the same time, China is reportedly weighing export curbs on 23 AI-related technologies, including chip designs and certain advanced AI models, to manage security and misuse risks, Bloomberg reported last week.

The Chinese AI lab has drawn attention from governments and regulators worldwide since its R1 reasoning model disrupted the AI sector in January 2025, causing a steep drop in Nvidia shares.

The US has accused DeepSeek and five other Chinese AI firms, including Alibaba and Moonshot AI, of employing large-scale model distillation to pull information from American AI systems, according to a September report.

Additionally, OpenAI and Anthropic have each expressed concerns about DeepSeek's reported use of model distillation, a method that leverages outputs from one AI model to train another. Earlier this year, OpenAI informed US lawmakers of suspected distillation attempts by DeepSeek, while Anthropic noted unauthorized efforts by DeepSeek and other Chinese AI labs to gather outputs from Claude.