| The evolution towards Beyond 5G (B5G)/6G systems is accelerating the emergence of distributed Edge–Cloud environments, where computation and intelligence span heterogeneous and dynamic infrastructures. While this enables latency-sensitive and data-intensive services, it also expands the attack surface, rendering traditional perimeter-based security insufficient. In this context, Artificial Intelligence (AI)-driven security is emerging as a key approach for enabling adaptive monitoring, intelligent threat detection, and automated response. This paper presents an integration-oriented perspective on AI-driven security in the Edge–Cloud continuum. It identifies the main security requirements and design dimensions, and analyses representative building blocks, including extended Berkeley Packet Filter (eBPF)-based monitoring, hardware-accelerated intrusion detection, federated intelligence, and privacy-preserving mechanisms. Based on these elements, the paper outlines a unified architectural framework that integrates telemetry collection, AI-driven detection, distributed learning, and trusted orchestration into an end-to-end security pipeline. The approach is further supported by insights from the ELASTIC and 6G-PATH projects, highlighting its applicability in realistic deployment scenarios. Finally, the paper discusses key challenges related to scalability, trust, and robustness in next-generation Edge–Cloud systems. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.