Rethinking Explanations through the Law: Systemic Explainability for Autonomous Vehicles in the European Union

“Artificial Intelligence is developing fast” and is central to legal, social, and ethical discourse for its potential to reshape modern living. Among the evolving applications of Artificial Intelligence (AI), this thesis focuses on autonomous vehicles and their critical impact over mobility paradigms in the European Union. Self-driving models are widely considered to be groundbreaking technology that is beneficial as much as it is disruptive. They are set to reshape mobility by, among other benefits, improving road safety, advancing sustainable transportation, reducing human error, and supporting traffic efficiency. However, their opaque, yet interactive and autonomous nature raises persistent technical, legal, ethical, and social challenges. Central among these are questions of risks in real-world driving environments, transparency in autonomous decision-making, utility and accessibility in automated transportation, as well as accountability in safety-critical scenarios, and the calibration of trust in human-computer interaction. Numerous studies show the benefits of autonomous vehicles, while addressing challenges and downsides, to support their development and on-road deployment. Amongst them, explainable artificial intelligence (XAI) research holds critical importance for the empowerment of humans involved in the value-chain of autonomous vehicles. Scholars argue that designing explainability in autonomous driving improves its safety, facilitates human-computer interaction, as well as directly improving transparency, fairness and interpretability. While explainability-oriented approaches are widely promoted as a response to the technical, social and ethical challenges of autonomous driving, their regulatory dimension remains underexplored. In particular, few studies analyze explainability in autonomous driving from a legal perspective. Existing scholarship in the juridical sciences underestimates the legal consequences of achieving, or failing to achieve, explainability in self-driving systems. Moreover, it rarely examines whether, and to what extent, regulation should actively support the transition towards explainable autonomous vehicles. This thesis addresses these gaps by advancing a regulatory framework for explainable autonomous driving. Its first objective is to articulate a legal perspective capable of realizing the systemic impact of explainability on the development and deployment of autonomous vehicles, including questions related to responsibility. Rather than considering XAI as a technical approach with incidental regulatory implications, this work conceptualizes explainability, as well as explanation-sharing, as a legally relevant design approach that shapes rights, obligations, risk allocation, and accountability structures throughout the autonomous driving value-chain. Second, the thesis argues that autonomous vehicles ought to be explainable through the law. It advances that explainability should be elevated to normative requirement for the design of autonomous vehicles in the Union, supported by targeted legal revisions. This combines with the argument above to show that XAI carries substantial legal consequences in the domain of autonomous driving, as the thesis challenges the prevailing tendency in the literature to treat explainability as external to legal analysis. Finally, this thesis proposes regulatory revisions aimed at facilitating the transition toward explainable autonomous driving. It argues that law should not be reactive to technological developments in the mobility sector and should hold a constitute role in steering innovation towards XAI-oriented solutions. In doing so, tis thesis aims to both support the systemic impact of explainability through law, and to expand it with legal perspectives embracing the relationship between XAI and themes of technology governance, accountability, liability, certification, and so forth. By doing so, this thesis contributes to the existing body of knowledge by identifying and tackling a structural gap at the intersection of law, explainable artificial intelligence, and autonomous vehicles. The thesis advances a law and technology approach that recognizes the critical and wide impact of explainability over autonomous driving, contributes to it by providing new legal perspectives, and offers regulatory ideas to support the transition toward explainable autonomous driving, realizing the systemic influence of XAI in the field. Ultimately, this thesis advocates for a future autonomous mobility paradigm in which self-driving vehicles are accountable, transparent, safe, and trustworthy through the law, while simultaneously avoiding one in which they remain opaque, dangerous and unpredictable due to regulatory gaps at the expense of explainability.