AI-Driven Cybersecurity Architecture for Saudi Smart Cities: Enhancing Threat Detection and Digital Resilience under Vision 2030
Abstract
-As part of Saudi Arabia’s smart -city agenda, Internet of Things devices, cloud platforms, cyber-physical systems, edge computing and artificial intelligence are increasingly being used to run urban services. This convergence poses a security challenge that cannot be addressed by traditional perimeter controls alone: attacks can spread across heterogeneous domains, exploit model and data dependencies, and disrupt critical services even when individual components are technically available. This paper presents an evidence-based architecture for AI-driven cybersecurity in smart cities in Saudi Arabia with a focus on threat detection and digital resilience within the framework of Vision 2030. This is a systematic integrative review of peer-reviewed research primarily published between 2020 and 2025 that synthesises evidence on intrusion detection, federated learning, edge intelligence, adversarial robustness, explainable AI, digital twins, smart-city risk, and Saudi cybersecurity readiness. The synthesis highlights the need for distributed detection at the data source, cross-domain correlation, explainable risk scoring, policy-constrained automation, and a resilience feedback loop connecting incident recovery and model assurance as key to effective architectures. Existing work shows strong potential for detection, but remains fragmented by domain, dataset and evaluation method; accuracy is often privileged over latency, recovery, privacy, resource cost and organisational governability. The paper thus proposes a layered architecture and resilience lifecycle that combines technical detection with the goals of governance and service continuity. The overall implications for Saudi deployments are that AI should be considered as an adaptive control capability in a governed socio-technical system, not just as a classifier. Future work should validate cross-city generalisation, adversarial robustness, federated collaboration, human supervision, and recovery-oriented metrics using Saudi data that are representative of operations.