Artificial Intelligence and the “Black Box” Dilemma in Fintech: Regulatory Perspectives in Azerbaijani Law

This article examines the legal, ethical and operational risks arising from the use of artificial intelligence (hereinafter AI) in critical processes within the fintech sector, such as credit and insurance risk assessment and fraud detection. The primary focus of the study centers on the “black box” phenomenon stemming from the opacity of deep learning algorithms’ internal logic, as well as the systemic biases and infringements upon citizens’ rights and legitimate interests arising in connection with this phenomenon. The paper comparatively analyzes proactive regulatory models formulated on the basis of the European Union’s Artificial Intelligence Act and General Data Protection Regulation standards. Although the role of Explainable AI in mitigating opacity is substantiated, it is emphasized that this technological approach alone is insufficient; consequently, the necessity of implementing “meaningful human control” and “human-in-the-loop” paradigms to prevent the adverse consequences of opacity is posited. Within this framework, it defines the role in decision-making of professionals responsible for the legal and factual soundness of decisions, such as credit and compliance officers and fraud investigators. Furthermore, throughout the research, Azerbaijan’s national development strategies and the regulatory instruments of the Central Bank are evaluated. Finally, specific normative proposals are put forward for national legislation aimed at strengthening algorithmic transparency and explainability requirements, as well as mandating human oversight for high-risk automated decisions.