Security, privacy, and agentic AI in a regulatory view: From definitions and distinctions to provisions and reflections

This paper examines how European Union regulations are evolving to address security and privacy concerns in AI systems, particularly as AI becomes more autonomous through 'agentic AI' that can make decisions independently. The authors analyzed 24 recent EU regulatory documents to clarify ambiguous definitions and explain how current rules apply to different types of AI systems. The paper provides guidance for policymakers and AI developers on ensuring compliance and responsible AI governance as these autonomous systems become more prevalent.