This review explores AI and ML-driven cybersecurity models for autonomous vehicles,
with emphasis on anomaly detection, adversarial robustness, and edge-computing-based defense systems.

A conceptual framework is proposed to enable real-time mitigation of cyberattacks in connected and autonomous transport systems.

Abstract The cybersecurity landscape for smart and autonomous vehicles is rapidly evolving, with new vulnerabilities emerging alongside advancements in communication, sensing, and AI integration. These systems rely heavily on complex algorithms and interconnected networks, making them increasingly susceptible to sophisticated cyber threats. Traditional intrusion detection systems (IDS) often fall short, lacking the adaptability to counter multi-variant attacks in real time. Emerging AI and machine learning (ML) techniques provide powerful alternatives, learning from new data and adapting to evolving threats. This paper conducts a comprehensive review of AI-driven approaches to securing autonomous vehicles and proposes a conceptual ensemble learning-based cyber threat intelligence framework. This framework aims not only to enhance the detection of complex attacks but also to integrate edge-level prevention capabilities, such as automated packet dropping, to proactively protect autonomous systems.