CNBC reported Tuesday that the U.S. Navy has asked personnel to avoid using any form of  DeepSeek, an allegedly low-cost high-performing Chinese-based AI model, due to potential security and ethical risks.

After the launch of its product, the Chinese startup DeepSeek topped the free download rankings in the U.S. Apple App Store, causing shockwaves in Wall Street, and technological circles. In addition to sparking widespread discussion about its potential effect in the AI computing race, questions were raised about the potential security risk posed by the app. U.S.-based media WIRED pointed out that DeepSeek routes user information directly back to China, potentially even more than TikTok in the past few years.

On Tuesday the U.S. Navy confirmed to CNBC that it had sent an email on January 24 warning colleagues to disable DeepSeek in any form. The warning detailed that employees were not to download, install, or use DeepSeek for any work-related or personal reasons, underscoring the new requirement as “extremely important”.

The announcement came the same week DeepSeek launched their DeepSeek-R1 model, which is said to have a lower construction cost, while maintaining performance comparable to that of industry leader OpenAI. As of Tuesday, DeepSeek still topped the App Store download list and has been downloaded more than 2 million times since January 15, with nearly 70% of the downloads in the past three days, according to market intelligence app AppMagic and U.S. media platform Semafor.

On Monday DeepSeek announced that it was temporarily suspending registration services to respond to “large-scale malicious attacks” on its services, although existing members were allowed to use the app as normal.