The US government has ordered civilian federal agencies to stop using artificial intelligence tools developed by Chinese startup DeepSeek, citing concerns that the technology could expose sensitive information to Chinese authorities amid growing competition between Washington and Beijing in the AI sector.
A notice issued by the US Department of Homeland Security on Monday directed agencies to remove DeepSeek products, along with third-party tools built using the company’s technology, from government information technology devices.
The department said DeepSeek’s models posed several security risks, including the potential transfer of confidential data to China when used by US government employees or contractors.
“DeepSeek’s models present multiple threat vectors,” a document attached to the notice stated.
The department warned that the company’s servers were located in China, where the legal system gives the government extensive authority to access private companies’ infrastructure and information.
The order applies to civilian government agencies under the department’s authority. It does not cover the military or intelligence services. The US Department of Defence did not respond to a request for comment from AFP.
AFP contacted DeepSeek for a response to the restrictions.
The decision marks another step in Washington’s efforts to limit the use of Chinese technology in sensitive government operations. US officials have increasingly raised concerns about data security and the potential for foreign access to information handled by artificial intelligence systems.
DeepSeek, one of China’s leading AI research companies, attracted international attention in January 2025 after releasing a low-cost chatbot that delivered performance comparable to leading systems developed by American technology companies.
Its emergence challenged assumptions that advanced AI development required the enormous spending and computing resources associated with US firms, intensifying competition over the cost and capabilities of AI models.
The company released its long-awaited V4 model in April and updated a faster, lighter version of the system this month.
Washington has also accused DeepSeek and other major Chinese AI developers of systematically taking capabilities from advanced models developed by US competitors. Such allegations have added to tensions over intellectual property, technology security and the future direction of the global AI industry.
Despite the rivalry, the two countries agreed in September to establish a communication channel for addressing AI-related incidents. The arrangement is intended to provide a means of discussing problems involving the rapidly developing technology.
Artificial intelligence is expected to feature in discussions when the US and Chinese leaders meet next month at a summit in Shenzhen, China.
The latest restriction highlights the growing overlap between AI competition and national security policy. As government agencies and private companies adopt increasingly sophisticated AI tools, questions over data storage, access rights and the jurisdiction governing technology providers have become central to decisions about which systems can be used in sensitive environments.
