The Stable Identity Assumption: Why Agentic IAM Is Solving the Wrong Problem

The emerging field of agentic identity and access management—across industry frameworks, national standards initiatives, and open protocols—rests on a shared foundational assumption: that agents are durable, nameable principals whose behavior is stable enough to govern through credential binding, role assignment, and audit trail attribution. This paper argues that assumption is structurally false for emergent agentic systems. Agents are not subjects. They are trajectories. What exists is not a persistent principal but a behavior envelope—a moving process whose identity is a runtime product of context, tools, prompt state, and prior agent interactions.