Artificial Intelligence for Financial Crime Detection: Economic Efficiency, Accountability, and Legal Responsibility
Year: 2026
Type: article
Abstract: Artificial intelligence (AI) is increasingly being applied to financial crime detection, offering potential efficiency gains over traditional rule-based systems while raising important questions concerning algorithmic accountability and criminal liability. This review examines 25 scholarly and institutional sources across law, criminology, computer science, and regulatory policy. It focuses on three areas: (1) the economic efficiency of AI in anti-money laundering (AML) and fraud detection; (2) criminal liability issues associated with false negatives, algorithmic opacity, and the due diligence responsibilities of system deployers; and (3) governance challenges relating to explainability, human oversight, and cross-jurisdictional regulatory harmonization. The review indicates that machine learning approaches offer potential advantages in identifying complex financial crime patterns and reducing false-positive alerts. However, the available evidence on overall cost-effectiveness and measurable crime reduction remains limited. False negatives represent a particularly significant challenge because undetected suspicious transactions may create regulatory, reputational, and legal risks. The analysis further highlights fragmented liability frameworks and the importance of due diligence in determining responsibility for AI-related failures. The review concludes that effective AI-based financial crime detection requires a human-machine collaborative governance approach supported by explainability, systematic false-negative monitoring, appropriate human oversight, and stronger international regulatory coordination.
Institutions: Sarbon University, Tashkent State University of Economics
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Keywords: artificial intelligence, financial crime detection, anti-money laundering, fraud detection, algorithmic accountability, criminal liability, false-negative results, algorithmic opacity, due diligence, explainability, human oversight, regulatory harmonization, cross-jurisdictional regulation, cost-effectiveness, human-machine collaboration, machine learning
Open Access status: hybrid
OpenAlex ID: W7234101609
