AI Act and new Product Liability Directive: The interplay between AI safety regulation and compensation

Abstract

The paper examines the interaction between the EU’s Artificial Intelligence Act (AIA) and the 2024 Product Liability Directive (PLD). Building on the premise that safety regulation and liability law constitute “two sides of the same coin,” it analyses the principal points of connection through which the AIA may shape liability under the PLD. These include defectiveness, PLD defences and liability for substantial modifications. With regard to the AIA, the analysis focuses on the mechanisms that are characteristic of EU product safety law, such as harmonised standards and conformity assessments. The paper demonstrates that the PLD contains several mechanisms capable of translating the AIA’s safety obligations into liability benchmarks, enabling both instruments to reinforce each other’s effectiveness. PLD claims may serve as a vehicle for private enforcement of the AIA, while the AIA can facilitate the assessment of liability. At the same time, the PLD’s traditional limitations – the product–service divide, restricted categories of compensable damage and exclusion of the user side of the value chain – constrain alignment precisely in areas where AI-related harms typically arise. These gaps are likely to be filled by diverging national liability regimes, increasing legal uncertainty and limiting the PLD’s capacity to provide a coherent EU-wide liability framework for AI-related harm.