The DHS notice warns that DeepSeek's models pose "multiple threat vectors" and that using them "may result in the transfer of sensitive data to the PRC" if government employees or vendors engage with the platform. The memo also says the models could produce "actionable intelligence on U.S. persons and/or government agencies."
The ban extends beyond DeepSeek's own app to cover third-party services and applications developed using the open-source model. A July DHS memo, made public on Monday, had already instructed agencies to monitor traffic to and from government websites for DeepSeek-related risks.
DeepSeek had already been restricted in other parts of the U.S. government. The Commerce Department and the Navy previously banned the chatbot on government devices, and the states of New York, Tennessee and Virginia have taken similar steps. The Federal Communications Commission is scheduled to vote on October 29 on a proposal to bar all Chinese laboratories from testing electronics for U.S. use.
The Chinese Embassy in Washington did not immediately comment.
DeepSeek, a Hangzhou-based startup, became widely known on January 27, 2025, when its open-source model, built for under $5 million, triggered a sharp drop in Nvidia's stock and erased roughly $1 trillion in value from U.S. technology stocks as it demonstrated capabilities comparable to America's best models at a fraction of the cost.
The company has continued to release updates. In April 2025 it released a preview of its V4-Pro model, and this week it released an updated version of its V4.1-Flash model aimed at efficiency. A House select committee on the Chinese Communist Party had previously alleged that DeepSeek secretly sends American user data to the Chinese government and may have used restricted Nvidia chips in its training.
Italy's data protection authority also restricted the chatbot in early 2025, calling DeepSeek's responses to privacy questions "totally insufficient."

