The applicable law in non–contractual liability arising from artificial intelligence systems within the european legal framework

Abstract

The expansion of artificial intelligence challenges the classical framework of tort law and compels private international law to reinterpret its categories. The difficulty of locating damage in digital environments, attributing liability without an identifiable human actor, and the substantive insufficiency of the applicable law undermine the effectiveness of traditional criteria such as the lex loci damni under the Rome II Regulation. The 1973 Hague Convention and Article 10.9 of the Spanish Civil Code complement this framework, yet they face limits when addressing harm caused by autonomous and opaque systems. Recent EU initiatives—namely the 2022 Draft Directives on defective products and AI liability, and the 2024 AI Act—seek to strengthen substantive protection through strict liability, reversal of the burden of proof, and harmonized standards of diligence. The challenge lies in articulating conflict–of–law techniques with substantive adequacy, ensuring that reparation does not depend on categories designed for an analog world.