Okta Announces New 'Auth0 for AI Agents' Features to Enable Secure Deployment

Okta, Inc. (Headquarters: San Francisco, USA), a provider of identity management services, has announced the latest features for 'Auth0 for AI Agents,' designed to eliminate the hurdles of custom development in authentication and authorization systems that have hindered the large-scale deployment of agentic applications.

Key New Features - **Auth for MCP**: The first enterprise-grade identity foundation to authenticate and authorize MCP (Model Context Protocol) clients. - **Agent as Principal**: Assigns unique identities to AI agents, allowing each action to be individually authorized and audited. - **On-Behalf-Of Token Exchange**: A mechanism for agents to securely exercise permissions on behalf of a user. - **Token Vault with Organizations Support**: Secure token management across complex organizational structures. - **FGA Permissions Index**: An authorization index that enables high-performance access control.

Background and Importance While approximately 80% of companies are experimenting with or adopting AI agents, only about one-third have achieved a 'complete and stable implementation.' Key gaps include treating AI agents as mere 'user extensions,' leading to excessive privileges and audit blind spots, as well as legacy access controls that hinder AI security and scalability.

Gareth Davies, Chief Product Officer (CPO) of Auth0, stated, 'With an independent identity foundation that securely connects AI agents to any tool, system, or provider, developers can focus on building superior experiences that drive revenue.'

Expected Impact These features will enable secure AI agent operations across industries, including highly regulated sectors like banking and healthcare for sensitive data access, retail for back-office system isolation, and legal investigations for strict case-level access restrictions.