Authors: Rohit Agnihotri.
Abstract: The rapid proliferation of cloud-native environments, decentralized workforces, and API-driven architectures has exposed critical limitations in conventional Identity and Access Management (IAM) systems. This article examines the emergence of Agentic Artificial Intelligence (AI)-driven IAM frameworks as a transformative paradigm for enterprise security. Drawing on empirical data from industry reports, peer-reviewed studies, and real-world deployment analyses, we evaluate the architecture, operational benefits, risk dimensions, and governance requirements of AI-powered IAM systems. Our findings indicate that agentic AI integration yields measurable improvements across breach detection (up to 87%), provisioning speed (82%), and compliance automation (91%), while simultaneously introducing novel challenges in auditability, model explainability, and adversarial robustness. We propose a multi-layered governance model and research agenda to guide responsible adoption.
DOI: https://doi.org/10.5281/zenodo.23209786
