Liability for Harm Caused by Artificial Intelligence: A Theoretical Framework of Fault, Causation and Responsibility Gaps in Common-Law Jurisdictions

Abstract

Artificial intelligence (AI) systems now diagnose patients, score borrowers, generate content and drive vehicles, and when they cause harm the common law of torts struggles to respond. Negligence presupposes an identifiable person whose unreasonable conduct caused the harm, but AI is autonomous, opaque, adaptive and produced through long value chains. Existing scholarship has examined particular reforms, especially in the European Union, but has offered no general framework for common-law jurisdictions that lack AI-specific legislation, and Nigeria has received almost no attention. This study addresses that gap. Its objectives were to identify the distinct gaps AI opens in common-law liability for harm, to evaluate the main remedial options against them, and to derive a reform framework for jurisdictions such as Nigeria. The study adopted a doctrinal, conceptual and comparative research design. It analysed primary legal sources, including leading common-law cases and statutes from the United Kingdom, the European Union and Nigeria, together with a purposive sample of 32 peer-reviewed journal articles published between 2023 and 2026, each verified against its Crossref publisher record, consulted through university library databases and screened for retractions. The analysis identifies three gaps: a fault gap, where harm occurs although no one acted unreasonably; a causation gap, where opacity prevents proof that a breach made a difference; and an attribution gap, where many contributors make it impossible to identify who is responsible. Four remedies were evaluated. Adapted negligence suits cases where human judgment remains central but leaves blameless harm uncompensated; strict product liability closes the fault gap but struggles to define defects in adaptive systems; presumptions with disclosure duties address opacity but collide with trade secrecy; and insurance or compensation funds secure compensation but may weaken incentives. The study argues for a layered regime and recommends that Nigerian courts apply negligence flexibly, that the legislature enact strict liability for defective AI products and services with disclosure duties and presumptions, and that compulsory insurance be piloted in high-risk sectors.