THE United States has ordered government agencies to stop using artificial intelligence tools developed by Chinese startup DeepSeek, citing security vulnerabilities and the risk of sensitive government data being transferred to China.
A notice issued by the Department of Homeland Security on Monday prohibits the use of DeepSeek products and third-party tools built using its technology, requiring them to be removed from government information technology devices.
The department can issue such directives to civilian agencies, but not to the military or intelligence services.
“DeepSeek models have various vulnerabilities that can be exploited as threats,” AFP reports today, citing a document accompanying the notice.
The department warned that using DeepSeek's AI tools could expose sensitive information to China if they were accessed by US government employees or contractors.
It said DeepSeek's servers were located in China, where the legal system gives the government broad powers to access the infrastructure of private companies.
The Pentagon did not respond to an AFP request for comment, while AFP has contacted DeepSeek for a response.
DeepSeek, one of China's leading AI laboratories, shook the industry in January 2025 by launching a high-performance, low-cost chatbot that challenged assumptions about US dominance in the sector.
The company launched its much-anticipated V4 model in April and released an updated, faster and lighter version this month, as Washington and Beijing compete for leadership in the rapidly expanding AI industry. - October 21, 2026