The United States Navy has issued a directive prohibiting its personnel from using DeepSeek, a rapidly growing AI chatbot application, citing security and ethical concerns related to the model's Chinese origins.
According to Navy officials, the restriction applies to both professional and personal use of the application. Personnel have been instructed not to download, install, or interact with the DeepSeek model in any capacity.
The ban comes as part of broader Department of Navy guidelines on generative AI usage, which also restricts other AI tools like ChatGPT. A Navy spokesperson clarified that while the recent communication specifically mentioned DeepSeek, it served as a reminder of existing policies against using publicly accessible AI systems for official work.
The directive gains particular relevance as DeepSeek, developed by a Hangzhou-based hedge fund, has attracted attention for creating an AI model using substantially fewer resources compared to major tech companies. The company's R1 model was trained using 671 billion parameters at a cost of $5.6 million, marking a stark contrast to the billions typically invested by leading tech organizations.
Adding to concerns, investigations reveal that DeepSeek collects extensive user data, including profile information, chat histories, and device details, storing this information on Chinese servers. The app also tracks keystroke patterns and usage across multiple devices.
The Navy's stance aligns with growing international scrutiny of DeepSeek. Italian and Irish authorities are investigating the company's data collection practices, while OpenAI and Microsoft are examining allegations that DeepSeek illegally accessed OpenAI's data output for training purposes.
This development reflects increasing caution among military and government institutions regarding the use of AI applications, particularly those with connections to foreign entities that might pose potential security risks.