No document available.
Abstract :
[en] This Chapter examines whether Aquilian liability can continue to provide a normative foundation for attributing fault in the age of Artificial Intelligence (AI). It argues that, while technological progress has repeatedly challenged longstanding conceptions of safety, causation and due diligence, previous industrial revolutions ultimately enabled law to consolidate sufficiently stable standards of safety and conduct. AI presents a more fundamental epistemic challenge: its opacity, adaptive capacity and causal indeterminacy make it difficult to identify, in advance, the risks that may materialize and to determine the precautions reasonably expected of providers and deployers. Against this backdrop, the Chapter critically examines the European Union’s approach to AI liability, focusing in particular on the proposed AI Liability Directive (AILD). It argues that the latter’s equation of fault with breach of legally prescribed duties risks reducing wrongfulness to formal unlawfulness and overlooking the limits of ex-ante risk regulation. Drawing on historical developments in technological liability and emerging AI litigation, the Chapter proposes a broader understanding of due diligence. For self-learning systems, diligence should constitute a continuing process of scrutiny: beyond preventing known or characteristic risks, providers and deployers should reasonably verify whether AI outputs remain sufficiently safe, reliable and trustworthy to support human decision-making. Aquilian fault should therefore encompass failures of ongoing verification, even where formal legal requirements have been satisfied.
