The global AI race is heating up, prominently showcasing HuggingFace as a key platform. Within just nine months, HuggingFace is expected to double its model offerings to 1.8 million, attracting public and private model providers. However, with this rapid expansion comes significant AI supply chain risks—vulnerabilities in software, models, and data. To address these threats, Cisco’s Foundation AI team has introduced Cerberus, an automated AI supply chain security guard monitoring HuggingFace continuously. Cerberus analyzes model updates for potential risks, generating standardized threat feeds for Cisco Security products. Recent integrations allow Cisco solutions to block downloads of compromised AI models, ensure compliance with open-source licenses, and prevent access to models from unapproved sources. By employing techniques such as metadata analysis and sandboxing, Cerberus enhances risk management. This proactive approach keeps cybersecurity practitioners ahead of emerging threats in the fast-paced AI landscape. Stay connected for more insights from Cisco Security.
Source link
Share
Read more