Critical National Infrastructures (CNIs)—including energy grids, water systems, transportation networks, and communication frameworks—are essential to modern society yet face escalating cybersecurity threats. This review paper comprehensively analyzes AI-driven approaches for Critical Infrastructure Protection (CIP). We begin by examining the reliability of CNIs and introduce established benchmarks for evaluating Large Language Models (LLMs) within cybersecurity contexts. Next, we explore core cybersecurity issues, focusing on trust, privacy, resilience, and securability in these vital systems. Building on this foundation, we assess the role of Generative AI and LLMs in enhancing CIP and present insights on applying Agentic AI for proactive defense mechanisms. Finally, we outline future directions to guide the integration of advanced AI methodologies into protecting critical infrastructures. Our paper provides a strategic roadmap for researchers and practitioners committed to fortifying national infrastructures against emerging cyber threats through this synthesis of current challenges, benchmarking strategies, and innovative AI applications.

Digital twin technology is crucial to the development of the sixth-generation (6G) Internet of Vehicles (IoV) as it allows the monitoring and assessment of the dynamic and complicated vehicular environment. However, 6G IoV networks have critical challenges in network security and computational efficiency, which need to be addressed. Existing digital twin technologies in 6G IoV networks often suffer from limitations such as reliance on static models and high computational demands, leading to unstable attack detection and inefficiencies. Their results for attack detection performance metrics, precision, detection rate, and F1-Score are insufficient for 6G IoV. Moreover, these systems concentrate all computational processes within the digital twin’s service layer, leading to inefficiencies. To address these challenges, we introduce a novel artificial intelligence (AI) enhanced digital twin framework designed to significantly improve 6G IoV network security and computational efficiency under dynamic conditions. Our framework employs an advanced feature engineering module that uses feature selection methods and stacked sparse autoencoders (ssAE) to reduce feature dimensions within the cyber twin layer, effectively distributing the overall computational load. It also utilises an online learning module which enables a network-aware attack detection mechanism for precise attack detection. The proposed solution exhibits a stable performance of around 98% success rate regarding attack detection metrics against two datasets. Specifically, our solution reduces system latency by 12%, energy consumption by 15%, RAM usage by 20%, and improves packet delivery rates by 6.1%. These findings underscore the potential of our framework to enhance the robustness and responsiveness of 6G IoV systems, offering a significant contribution to vehicular network security and management.

Digital Twin-as-a-Service (DTaaS) has emerged as a transformative paradigm to deliver modular, scalable, and intelligent digital twin (DT) functionalities. Regarding this, several studies have investigated DT platforms and their enabling technologies. However, their focus mostly remains domain-specific, thereby limiting the potential for generalisation across diverse contexts. Furthermore, these studies lack a systematic framework to address both architectural elements and their performance evaluation required for an end-to-end DTaaS platform. To bridge this gap, in this survey, we present a comprehensive component-based taxonomy for DTaaS platforms, categorised into core and supportive components. As the core components, we review Data, Model, Service, and Communication, whereas the supportive components include Infrastructure, Orchestration, and Security elements. For each component, we analyse the state-of-the-art methods, enabling technologies, and associated design trade-offs. We further introduce a metric-driven perspective to design measurable and context-aware DTaaS platforms. In this regard, we present several DT performance metrics, including twinning rate, twin fidelity, and age-of-twin and map them to DTaaS components. To demonstrate the applicability of our framework, we present a case study based on our own DTaaS platform. We perform two metric-driven implementations, (i) Radio Access Network (RAN) management and (ii) Distributed Denial of Service (DDoS) attack detection by exemplifying how DT-metrics can be integrated with the DTaaS components. Also, we outline key open issues for context-aware, interoperable, and secure DTaaS platforms. By jointly addressing an end-to-end architectural design and performance metrics, this survey demonstrates a foundational reference for the structured development, evaluation, and operation of DTaaS platforms.

This paper delves into enhancing cybersecurity training efficacy through an in-depth examination of gamified learning, behavioural strategies, and the deployment of digital twins. It identifies the critical role of behavioural strategies in bolstering cybersecurity defences by influencing human behaviour. The study explores the benefits of gamified learning in engaging participants and improving knowledge acquisition, alongside applying digital twins and cyber ranges in offering practical, hands-on experience. By analysing these methodologies, the paper aims to bridge the gap between theoretical knowledge and real-world application, encouraging the researchers to work on a forward-looking approach to cybersecurity training using these innovative methodologies that are both effective and engaging. Our findings suggest that by creating an immersive, interactive learning environment, it is possible to enhance the cybersecurity competencies of individuals, making them better prepared to navigate the complexities of the digital age. This paper contributes to cybersecurity training by offering insights using innovative approaches for effective training programs that are both engaging and informative, ultimately aiming to bolster organisations’ cybersecurity posture.

The rise of sixth-generation (6G) wireless technologies is expected to transform local connectivity through ultra-low latency communication, autonomous network management, and massive device integration. Among these, 6G Local Area Networks (LANs) are emerging as key enablers for smart homes and industrial automation, offering decentralised and intelligent control at the network edge. However, it introduces critical security requirements. Traditional intrusion detection methods, based on static rules and signatures, are limited in addressing dynamic device interactions, strict latency constraints, and increasingly stealthy cyberattacks expected in 6G LANs. To overcome these limitations, this work introduces a graph-based security framework using digital twins (DTs) designed for 6G LANs. The proposed solution integrates three core components. First, a graph-based anomaly detection module models the real-time network structure using GraphSAGE and Graph Attention Networks (GAT) to enable efficient and context-aware threat identification. Second, an explainability module based on Shapley Additive Explanations (SHAP) provides transparent decision-making and introduces a confidence-aware mechanism to autonomously or collaboratively handle detected threats. Third, a self-healing module applies mitigation actions dynamically, enabling fast, adaptive, and explainable responses while allowing human oversight when needed. The effectiveness of the framework is validated through extensive experiments on a 6G LAN testbed and an external dataset. Results show that our solution reduces end-to-end detection latency by up to 63%, lowers mitigation reaction time by 55%, and decreases processing overhead by 20%, while improving detection accuracy by up to 25% compared to conventional methods. These results highlight the practical benefits of using DTs for real-time and explainable 6G LAN security. Our work offers a scalable and locally deployable solution that meets the performance and transparency needs of emerging autonomous networks.