The U.S. Department of Homeland Security has issued a directive requiring civilian government agencies to remove the DeepSeek AI chatbot from their systems. Agencies are also prohibited from purchasing the software or any third-party services built upon it. The department cited concerns over potential data transfers to China and the risk of generating actionable intelligence on U.S. citizens and government entities.
This federal action follows earlier restrictions implemented by the U.S. Department of Commerce and the Navy, as well as state-level bans in New York, Tennessee, and Virginia. A July memo had previously instructed agencies to monitor network traffic for risks associated with the platform. These measures coincide with broader federal efforts to limit Chinese technology, including an upcoming FCC vote on October 29 regarding the testing of electronics.
DeepSeek, a startup based in Hangzhou, gained significant attention on January 27, 2025, following the release of its open-source model. Reports from a House select committee allege that the company may have utilized approximately 60,000 Nvidia chips, including 20,000 units subject to U.S. export controls. The committee further claimed that the platform may have secretly transmitted American user data to the Chinese government.
Despite these regulatory challenges, the company continues to release new software. In April 2025, it launched the V4-Pro model, priced 98% lower than OpenAI’s GPT-5.5 Pro. The company also recently updated its V4.1-Flash model. Internationally, the Italian data protection authority has also ordered DeepSeek to block its chatbot, citing insufficient responses to privacy inquiries.
