Pentagon Staff Continue Using DeepSeek's AI Model Amid Security Concerns

U.S. military personnel have reportedly been utilizing an early version of the AI model from the Chinese startup DeepSeek for several months.

According to Bloomberg, Pentagon employees began downloading this generative artificial intelligence model onto their workstations in the fall of 2024. The Pentagon's IT experts only partially blocked the use of DeepSeek after it gained significant popularity.

Recently, DeepSeek introduced an open-source AI model named R1, which its developers claim outperforms several top products from U.S. companies, including OpenAI’s flagship models in certain instances. The model's open accessibility, allowing users to download and run it on personal servers for free, has energized the open-source community and led to notable declines in U.S. tech stock prices.

The Defense Information Systems Agency (DISA), responsible for the Pentagon’s IT networks, took action to block access to DeepSeek's website late on a Tuesday. This came after many defense employees had already used the latest AI chatbot for at least two days, as reported by Bloomberg sources.

DISA experts are now working to assess how extensively employees have utilized DeepSeek through web browsers. DeepSeek's privacy policy indicates that user data is stored on servers in China and is subject to Chinese laws when accessed via its website or popular apps found on the Apple App Store and Google Play.

Despite the restrictions, thousands of Pentagon staff are reportedly still using DeepSeek through Ask Sage, an authorized platform that hosts various AI models on U.S.-based servers.

Some military personnel downloaded an earlier version of DeepSeek as early as the fall of 2024. At that time, the connection to China was allegedly “not evident” to Defense Department security teams, and the downloads did not raise immediate alarms.

The increasing attention on DeepSeek has led the U.S. military to seek out and erase code from Chatbots originating in China that might be on employees' computers. The Navy has outright banned the use of DeepSeek due to security and ethical concerns associated with its origins. Meanwhile, the Air Force has not provided specific guidance regarding DeepSeek but prohibits using sensitive public information in commercial generative AI systems without proper authorization.

OpenAI’s CEO and co-founder, Sam Altman, acknowledged DeepSeek R1 as an “impressive model, particularly regarding what they deliver for the price.” DeepSeek claims to have developed its competitive AI model without utilizing advanced U.S. chips, with a reported investment of just $6 million.

Former U.S. President Donald Trump, who prioritized AI development during his administration, referred to DeepSeek’s launch as a “wake-up call” for the American AI sector. He recently announced the Stargate initiative, aimed at establishing data centers across the U.S. to support upcoming AI projects with a targeted investment of up to $500 billion.

Pentagon, AI, DeepSeek