Okta Inc. (NASDAQ: OKTA) announced a security framework for enterprises managing AI agents and introduced Okta for AI Agents, a platform scheduled for general availability on April 30, 2026.

The framework addresses three questions for organizations: where AI agents are located, what they can connect to, and what actions they can perform. The platform aims to help companies discover both sanctioned and unsanctioned AI agents, control their access to resources, and revoke access when needed.

According to research cited by Okta, 88% of organizations report suspected or confirmed AI agent security incidents, while only 22% treat AI agents as independent entities requiring identity management.

The Okta for AI Agents platform includes several components. Agent integrations in the Okta Integration Network will support platforms including Boomi, DataRobot, and Google Vertex AI. The system can detect unauthorized "shadow agents" created by employees and register all agents as managed identities.

The platform features an Agent Gateway that serves as a control point for AI agent access to resources, along with credential management and API access controls. A "Universal Logout" function can immediately revoke all access tokens for agents that deviate from intended behavior.

"AI agents are evolving faster than any software before them, making traditional security models obsolete," said Ric Smith, President of Products & Technology at Okta.

The announcement comes as AI agents become more capable, with some able to execute terminal commands, access file systems, and perform complex workflows autonomously. Okta is expanding its Universal Directory to treat AI agents as distinct, non-human identities requiring lifecycle management.

The information is based on a company press release.