The useful question is what changes for users, developers or buyers, and whether the announcement stays industry context or becomes something people can actually use.

Okta has launched new governance capabilities for AI agents, enabling organizations to certify agent connections, enforce access revocations, and delegate management to scoped admin roles. These features reduce risk by limiting Super Admin access and allow admins to configure agent integrations and monitor activity. The updates address compliance needs in regulated sectors, including FedRAMP High, FedRAMP Moderate, and HIPAA environments.

A new SKU unlocks Okta for AI Agents capabilities for customers in regulated environments, though it excludes shadow AI agent discovery and secure connections to service accounts and secrets. Okta is actively working to authorize these features. The platform also introduces an identity layer for multi-agent architectures, providing visibility and audibility for agent-to-agent handoffs in complex workflows.

Okta’s latest offering includes fine-grained authorization for AI agents, moving beyond static roles to continuous evaluation based on real-time context. This enables customers to define, audit, and secure AI agents as distinct machine identities with strict runtime permissions. The system replaces static credentials with restricted, short-lived tokens to prevent cross-tenant data leaks and supports third-party API integrations.

The platform now supports importing AI agents built on Amazon Bedrock AgentCore, allowing direct governance within Okta for AI Agents. It integrates with non-Okta identity providers, offering a purpose-built agent identity platform without requiring changes to existing human identity infrastructure. Additionally, Okta optimizes relationship-based access control by indexing permission paths, enabling AI agents to verify document-level user authorization in milliseconds.