US Defense Department employees connected their work computers to Chinese servers to access DeepSeek’s new AI chatbot for at least two days before the Pentagon moved to shut off access

Employees of the U.S. Defense Department accessed DeepSeek’s AI chatbot using their work computers, inadvertently connecting to Chinese servers for at least two days before the Pentagon intervened to cut off access, according to a defense official familiar with the situation.

The Defense Information Systems Agency, which manages the Pentagon’s IT networks, moved to block access to the Chinese startup’s website late Tuesday, the official and another source said. Both spoke on condition of anonymity, as the information is not publicly available.

The restriction followed concerns from defense officials about Pentagon employees using the tool, one source explained. DeepSeek’s privacy policy states that user data is stored on Chinese servers and governed under Chinese law, raising security risks.

By Wednesday, some Pentagon computers displayed a “Website Blocked” message citing operational reasons, but access to DeepSeek remained available on others, according to the defense official and internal correspondence reviewed by Bloomberg News.

The Defense Information Systems Agency declined to comment.

DeepSeek’s AI model, reportedly developed for under $6 million, made global waves earlier this week, drawing attention to the vast sums—hundreds of billions of dollars—U.S. tech firms spend on AI infrastructure. The model soared to the top of the Apple Store charts, with figures like Marc Andreessen praising its capabilities. However, concerns have escalated over the chatbot’s security implications, particularly regarding data storage in China.

Pentagon IT experts are still assessing how extensively employees engaged with DeepSeek’s system via web browsers, the defense official said.

Military personnel had already begun downloading an earlier version of DeepSeek’s code on workstations in fall 2024, a source familiar with the matter revealed. At the time, Defense Department security teams did not flag the activity, as the tool’s connection to China was not immediately apparent.

With the latest DeepSeek release sparking widespread interest, some military branches are now working to locate and remove chatbot-related code from employees’ devices, the source added.

Meanwhile, thousands of Defense Department personnel are using DeepSeek through Ask Sage, an authorized software platform that does not directly link to Chinese servers, according to Nicolas Chaillan, Ask Sage’s founder and CEO.

The military’s approach to DeepSeek varies by service. The Navy, for instance, formally banned its use on Friday, citing security and ethical concerns related to the model’s origins, according to CNBC.

Navy spokesperson Lieutenant Commander Lauren Chatmas confirmed that the branch already has guidance restricting the use of open-source AI tools for official work. Recent internal communications explicitly referenced DeepSeek in relation to that policy, she added.