Reforming the EU Civil Liability Framework Applicable to Artificial Intelligence and Other Emerging Digital Technologies: The Proposed AI Liability Directive

This chapter provides a thorough and systematic analysis of the European Commission’s proposal for a Directive on adapting non-contractual civil liability rules to artificial intelligence (the ‘proposed AI Liability Directive’). Current fault-based national liability rules are not fit to deal with compensation claims for harm caused by AI-enabled products/services. Indeed, under such rules, victims need to prove a wrongful action/omission of a person that caused the damage. In other words, the burden is on the victim to identify the responsible party and to prove the fault, damage, and causal link. In the absence of targeted measures to ease the burden of proof, the specific characteristics of certain AI-systems could make it extremely difficult and/or prohibitively expensive for victims to identify and prove the fault of a potentially liable person as well as the causal link between that fault and the damage suffered. Tracing the victim’s loss back to the defendant may be challenging if the suspected cause of that loss is related to the operation of an AI system, due to the increasingly autonomous nature of some AI systems and their peculiar characteristics. The proposed AI Liability Directive harmonises targeted aspects of the Member States’ existing civil liability rules applicable to AI-systems and its aim is to complement and modernise the EU liability framework by introducing specific rules for damages caused by AI systems. It attempts to address the AI-specific problems by improving the rights of victims who have suffered any damage—as admissible under national law—due to wrongful use of artificial intelligence. In particular, the proposed AI Liability Directive introduces two key measures: First , the ‘presumption of causality’, which relieves victims from having to explain in detail how a specific fault or omission caused the damage. If victims can show that a party was at fault for not complying with a relevant obligation and that a causal link with the AI performance is reasonably likely, the court can presume that this non-compliance caused the damage. The liable party, however, can rebut this presumption by proving that another cause was responsible for the damage. Second , it assists victims in accessing relevant evidence from companies or suppliers when dealing with high-risk AI systems. Victims can request the court to order the disclosure of information about high-risk AI systems, facilitating the identification of the liable party and the understanding of the fault. This disclosure is subject to safeguards to protect sensitive information, including trade secrets. Any type of victim (natural or legal person) can bring a claim for damages against any person for fault that influenced the AI system which caused the damage. The victim can be compensated if harmed by the fault or omission of a provider, developer or user of AI resulting in a damage covered by national law (e.g., health, property, including damage caused by safety issues or resulting from unlawful discrimination or breach of fundamental rights like privacy etc.). The AI Liability Directive, once ratified, is expected to facilitate compensation for victims—whether individuals or businesses—for damages caused by an AI system’s output (certain faults or omissions), by enhancing their ability to file successful claims against any party whose fault influenced the AI system causing the damage.