Liability for damage caused by autonomous ships in light of the EU framework on artificial intelligence

With the advent of remotely controlled and autonomous ships, the IMO has undertaken an effort to define a regulatory framework for maritime autonomous surface ships (MASS). In parallel, the EU has made some attempts to produce a framework on civil liability for artificial intelligence (AI). Autonomous and remotely controlled navigation disrupts two foundational premises of maritime law: that ship must sail with a crew and under the authority of a master. Such changes promise to have a deep impact, not only in practice in the maritime trade, but also on legal rules, including those on liability for damages. This paper traces the legislative evolution from the European Parliament’s 2020 Resolution envisaging a strict liability regime for AI systems, to the 2022 AI Liability Directive proposal and the 2024 AI Act, which adopt a procedural and fault-based approach designed to alleviate evidentiary burdens of potential victims of damages caused by AI systems. The latest approach of the EU contrasts sharply with the strict liability and compulsory insurance models underpinning classic maritime conventions such as the CLC 1992, the Bunkers 2001, or the HNS 1996, which ensure effective victim compensation for technologically induced risks. The article argues that the shipowner’s traditional centrality in maritime liability should be probably preserved in any future rules, as it has reasonably proved to sufficiently protect victims and effectively provide for a fair risk distribution and damage socialization through liability insurance.