Identity Management for Agentic AI: The new frontier of authorization, authentication, and security for an AI agent world
By: Tobin South , Subramanya Nagabhushanaradhya , Ayesha Dissanayaka and more
Potential Business Impact:
Secures AI agents and their access to information.
The rapid rise of AI agents presents urgent challenges in authentication, authorization, and identity management. Current agent-centric protocols (like MCP) highlight the demand for clarified best practices in authentication and authorization. Looking ahead, ambitions for highly autonomous agents raise complex long-term questions regarding scalable access control, agent-centric identities, AI workload differentiation, and delegated authority. This OpenID Foundation whitepaper is for stakeholders at the intersection of AI agents and access management. It outlines the resources already available for securing today's agents and presents a strategic agenda to address the foundational authentication, authorization, and identity problems pivotal for tomorrow's widespread autonomous systems.
Similar Papers
Authenticated Delegation and Authorized AI Agents
Computers and Society
Lets you control what AI helpers can do.
Binding Agent ID: Unleashing the Power of AI Agents with accountability and credibility
Networking and Internet Architecture
Keeps AI from being misused by proving who's in charge.
Systems Security Foundations for Agentic Computing
Cryptography and Security
Makes AI agents safer from hackers.