Foundation AI Enhances AI Security in Collaboration with Hugging Face

Hugging Face is rapidly expanding, adding a new AI model every 7 seconds and now hosting nearly 1.9 million models globally. This growth highlights the urgent need for enhanced AI supply chain security due to risks like vulnerable software dependencies and malicious model files. To address these challenges, Cisco’s Foundation AI team has partnered with Hugging Face, integrating comprehensive malware scanning capabilities using the updated ClamAV engine. This collaboration provides rigorous model vetting, early vulnerability detection, and shared threat intelligence. Notably, ClamAV has introduced features to detect malicious code in AI models and is the sole antivirus engine focused on AI risks within VirusTotal. Additionally, Cisco’s Cerberus system ensures real-time inspection of models entering Hugging Face, enhancing security measures across the AI ecosystem. These initiatives empower organizations and individuals to confidently adopt AI while safeguarding against supply chain hazards, illustrating a commitment to robust cybersecurity in the AI landscape.

Source link