The United States of America Navy has officially banned the use of China’s DeepSeek AI across all operations.
In a memo, the Navy informed personnel that the generative AI model must not be used “in any capacity,” citing serious security and ethical risks tied to its Chinese origins.
The directive applies to both work-related and personal use.
This strict decision comes as DeepSeek’s AI technology crashes global financial markets and triggers debate about the future of AI and US dominance in it.
But the Navy’s warning, which was distributed to all operational personnel, actually came days before the markets went ballistic over DeepSeek’s latest model, R1, which rivals tech from US companies like OpenAI.
The memo, distributed under the Operational Navy (OpNav) system, left no room for ambiguity.
“We would like to bring to your attention a critical update regarding a new AI model called DeepSeek. It is imperative that all personnel refrain from downloading, installing, or using DeepSeek’s AI in any capacity. This includes work-related tasks and personal use.” it reads in part.
The message was signed off by the Naval Air Warfare Center Division Cyber Workforce Manager, showing the high-level concerns over DeepSeek.
Unsurprisingly, the concern comes mainly from DeepSeek’s status as an open-source model, meaning it is accessible to developers worldwide, including those operating in high-risk environments. America thinks that’s a problem, likely because they are not familiar with open-source at all.
DeepSeek claims the R1 was built in just two months with a modest $6 million budget. The app has since skyrocketed to the top of Apple’s App Store, overtaking OpenAI’s ChatGPT in popularity.
The Pentagon has gotten more and more wary of the AI’s rapid adoption, particularly given its origins in a country under heavy US tech sanctions.
The ban also comes as DeepSeek temporarily restricted user registrations earlier this week, citing “large-scale malicious attacks” on its systems. Operations resumed shortly afterward, but the incident heightened concerns over the security vulnerabilities of open-source.
“DeepSeek is a wake-up call,” President Donald Trump, who took office just last week, declared during a press briefing on Jan. 27. “America’s tech companies need to step up. This is not the time to get comfortable.”
David Sacks, a venture capitalist and Trump’s newly appointed AI and crypto czar, also commented on the situation, saying, “DeepSeek R1 shows just how competitive the AI race has become. I’m confident in the US, but we cannot afford complacency.”
The Trump administration’s first major AI deal was last week’s announcement of Stargate, a joint venture between OpenAI, Oracle, and SoftBank. The project will funnel over $500 billion into AI infrastructure in a mission to solidify America’s AI dominance.
DeepSeek has done a lot of damage though. On Monday, Nvidia suffered its worst trading day since March 2020, with shares plummeting by 16.9%. Broadcom followed closely, with its stock falling 17%. Together, these two giants lost $800 billion in market value, making the worst kind of history.
Micron Technology and Arm Holdings dropped over 11% and 10%, respectively, while AMD and Broadcom saw losses of 6% and 17%. Even power infrastructure companies tied to AI development, like Constellation Energy and Vistra, faced massive declines of up to 28%.
Nvidia did come out and acknowledge DeepSeek’s innovation that very same day.
In a statement, the chipmaker said, “DeepSeek is an excellent AI advancement and a perfect example of Test Time Scaling. Their work shows how new models can be created using widely available models and export-compliant compute.”