Poison Control: Unifying Software and Content... Anmol Krishan Sachdeva & Ankit Kotnala, Google

This session from Google engineers introduces the Dual-Provenance Pattern for securing AI workloads on Kubernetes, using Kyverno as a Policy-as-Code gatekeeper to enforce cryptographic signatures on both software artifacts and AI model/content provenance via C2PA standards. It addresses the critical blind spot where AI models and training datasets remain unverified while container runtimes are secured via SBOMs and binary authorization. Ideal for platform engineers and security teams building zero-trust defense for AI pipelines.