Abstract
<title>Abstract</title> <p>This article examines artificial intelligence (AI) in anti-money laundering and countering the financing of terrorism (AML/CFT) as a criminological, legal and regulatory control problem, rather than a financial-technology efficiency exercise. It explores how AI reshapes opportunities for money laundering and terrorist financing, affects AML/CFT legal processes, and dictates the governance conditions required for AI-enabled systems in compliance and investigative decisions. The article adopts a structured criminological evidence map and critical narrative synthesis. Sources include academic databases, legal instruments, policy reports and technical studies, coded by AML/CFT function, risk, design, jurisdiction and policy implication. While Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) 2020 informed the reporting logic, the review is not PRISMA-compliant due to the heterogeneous nature of the materials. Findings reveal that evidence is strongest for machine-learning, graph-analytics and blockchain-analytics, but weaker for generative-AI-assisted reporting, CFT-specific detection and independent live deployment. The review distinguishes static tabular credit-risk models from dynamic relational AML/CFT problems, narrows claims about synthetic data, and connects model governance to cyber-resilience. While AI improves alert prioritisation and network detection, due diligence and screening, criminal actors use AI to scale identity fraud, deepfakes, mule recruitment and obfuscation. CFT use requires caution, as humanitarian operations and remittances create risks of under-detection and discriminatory over-enforcement. The article reconceptualises AI-enabled AML/CFT as a co-evolutionary dual-use control problem involving regulators, intermediaries and criminal enterprises. It develops a multilevel governance framework connecting opportunity reduction, legal accountability, evidence quality, supervisory capacity, rights protection and disruption of illicit financial flows.</p>