Okta Acquires Permiso to Secure AI Agents After Login

Giving an AI agent permission to enter a system is only the first security decision. The harder question is what it does once inside.

What happened

Okta has signed a definitive agreement to acquire Permiso Security, a cloud-native platform that detects suspicious behaviour across human users, service accounts and AI agents in multi-cloud environments. Financial terms were not disclosed.

Okta is best known for authentication and identity management: confirming who or what is requesting access and applying the right permissions. Permiso focuses further into the session. It monitors behaviour after access has been granted, looking for signals such as excessive privileges, unusual tool use, policy violations and activity that could create a large security impact.

Permiso’s technology draws on more than 2,500 research-based signals across over 70 identity partners. Okta plans to integrate those capabilities into its broader identity-security platform.

Why it matters

Traditional identity systems were designed mainly around employees logging into applications. Companies now also manage service accounts, automated processes and AI agents that can take actions across several tools without continuous human supervision.

An agent may be correctly authenticated and still behave dangerously because of a compromised account, a malicious instruction or permissions that are too broad. Security therefore needs to cover runtime behaviour, not just the login screen.

The bigger picture

AI agents are turning identity security into an ongoing control problem. The market is expanding from “who can enter?” to “what is this identity doing, and should it be stopped?” Okta’s acquisition suggests that authentication, privilege management and behavioural monitoring are starting to merge into one security layer for humans and machines.