Abstract The Connected and Autonomous Vehicle (CAV) is an emerging mobility technology that may hold a paradigm-changing potential for the future of transport policy and planning. Despite a wealth of likely benefits that have made their eventual launch inescapable, CAVs may also be a source of unprecedented disruption for tomorrow's travel eco-systems because of their vulnerability to cyber-threats, hacking and misinformation. CAVs manipulated by users, traffic controllers or third parties may act in deceitful ways. This scene-setting work introduces the deceitful CAV, a vehicle that operates in a deceitful manner towards routing and control functionality for ‘selfish’ or malicious purposes and contextualises its diverse expressions and dimensions. It specifically offers a systematic taxonomy of eight distinctive deceitful behaviours namely: suppression/camouflage, overloading, mistake, substitution, target conditioning, repackaging capability signatures, amplification and reinforcing impression. These as exemplified by their most common attack forms (i.e., starvation, denial-of-service, session hijacking, man-in-the-middle, poisoning, masquerading, flooding and spoofing) are then benchmarked against five key dimensions referring to time frame (short to long duration), engagement (localised to systemic), urban traffic controller infrastructure (single to multiple components), scale (low to high), and impact (low to high). We then suggest mitigation strategies to protect CAV technology against these dangers. These span from purely technological measures referring to the machine-centric triad of vehicles, communication, and control system including adversarial training, heuristic decision algorithms and weighted voting mechanisms to human factor measures that focus on education, training, awareness enhancement, licensing and legislation initiatives that will enable users and controllers to prevent, control or report deceitful activities.

Highlights • CAVs manipulated by users, traffic controllers or third parties can become deceitful. • The paper introduces a taxonomy of deceitful behaviours: suppression/camouflage, overloading, mistake, substitution, target conditioning, repackaging capability signatures, amplification and reinforcing impression. • These are benchmarked against five key dimensions referring to time, engagement, urban traffic controller infrastructure, scale and impact. • Anti-deception mechanisms consist of technological measures including adversarial training, heuristic decision algorithms and weighted voting mechanisms and human factor measures that focus on education, training, awareness, licensing and legislation. • This is a scene-setting work positioning a diverse traffic control problem and opening up pro-actively an interdisciplinary dialogue.