Concerns Raised Over Exposed Ollama APIs and DeepSeek Models

🔒🧩 Exposed Ollama APIs raise concerns over AI data security and DeepSeek adoption. The rapid growth of AI tools, particularly through the Ollama framework, has led to significant adoption but also increased risks of data leaks due to misconfigurations. Analysis reveals around 7,000 IPs with exposed Ollama APIs, with 700 running DeepSeek models, predominantly in China and the U.S. The U.S. Navy and other organizations have banned DeepSeek over security concerns, highlighting the potential for unauthorized access and exploitation. As hobbyists deploy these models on less secure networks, the need for robust security measures and audits of exposed APIs becomes critical to mitigate risks associated with AI data leakage.

Source

Original