Civil Liability for AI-Supported Banking Decisions: A Risk-Graduated Model under EU and Algerian Law

Abstract

The increasing use of artificial intelligence (AI) in banking raises a distinctive problem of civil liability. AI-supported credit assessment, credit scoring, fraud detection, transaction monitoring and risk management may materially affect customers’ financial and legal positions, while the causes of a decision may be distributed among banks, software developers, technology providers and data suppliers. This article examines whether traditional civil-liability rules can allocate responsibility for harm caused by AI-supported banking decisions and whether the European Union’s risk-based regulatory approach can inform the development of a more coherent Algerian framework.

The analysis focuses on Regulation (EU) 2024/1689 (the AI Act), particularly its treatment of high-risk AI systems used for creditworthiness assessment and credit scoring, together with the Algerian Civil Code, Law No 23-09 on Monetary and Banking Law, and recent Bank of Algeria regulations concerning customer protection and digital banking.

The article argues that the AI Act is not a comprehensive civil-liability regime. Its ex ante requirements on risk management, data governance, documentation, record-keeping, transparency, accuracy, robustness and human oversight may nevertheless provide evidence relevant to the standard of care in subsequent liability disputes. Against this background, the article develops a risk-graduated model based on five connected criteria: algorithmic risk, effective control over the source of risk, duty of care, legally cognisable harm, and causation.

The model seeks to avoid both automatic attribution of liability to the bank and unjustified transfer of liability to developers or technology providers. It concludes that Algeria should adapt, rather than simply transplant, European preventive and evidentiary safeguards to its domestic civil-law and banking framework.