Designing Identity for the Agentic Enterprise: The Okta AI Identity Summit

This blog post summarizes key themes from the Okta AI Identity Summit, arguing that identity must evolve from traditional access control to a runtime control plane that governs agentic AI. Speakers emphasized discovery of shadow agents, lifecycle-aware identity, action-level authorization, accountability models, human-in-the-loop controls, and operational controls (registries, kill switches, continuous access reviews) to manage the risks of agents acting with delegated authority.