Security experts have found over 100 harmful AI models on the Hugging Face platform, capable of hacking into users' systems. These models use a method that allows attackers to control victims' machines secretly, posing risks of data breaches and espionage globally. One particular model connects to an IP address in Korea, hinting at potential widespread threats. The discovery highlights the dangers within open-source repositories and the evolving tactics, like the generative AI worm Morris II, that can spread malware and steal data across systems by exploiting AI models' capabilities. This situation underlines the critical need for better security measures in AI technologies.

Over 100 Malicious AI/ML Models Found on Hugging Face Platform