Zero Trust in the Age of AI: Why the Classic Model Isn’t Enough Anymore

**Executive summary:** This article argues that traditional Zero Trust architectures—designed for human users—are insufficient for AI agents and the surging number of machine identities; it outlines AI-specific risks (autonomous multi-hop workflows, permission amplification, prompt injection) and prescribes pragmatic controls including per-agent workload identities, short-lived and just-in-time permissions, input/output validation, agent-specific behavioral baselines, and human approval for high-risk actions.