E195: Taking on the New AI Attack Surface With Manifold: Runtime, Skills & Supply Chains

AI security is undergoing a fundamental shift from securing static model responses to protecting dynamic, agent-driven actions at the endpoint. This transition requires moving beyond traditional guardrail-based approaches toward comprehensive runtime detection and response. Manifold Security addresses this by prioritizing discovery and inventory management, ensuring organizations understand the full supply chain of their AI agents, including external skills and plugins. The open-source project Manifest serves as a foundational tool for this effort, providing a common ground truth to identify malicious artifacts without generating excessive noise. By focusing on lineage and author trust scores rather than simple classification, this approach enables security teams to govern AI adoption effectively. Successful AI security strategies now demand a balance between rapid innovation and enterprise-ready operationalization, moving away from theoretical models toward practical, scalable runtime protection.