Abstract Artificial intelligence is reshaping identity verification and risk assessment in banking. This shift is occurring amid escalating threats from synthetic identities, deepfake-enabled impersonation, and automated fraud. This narrative literature review synthesises research on responsible deployment in Know Your Customer (KYC) processes, biometric and document verification, machine-learning fraud detection, and the governance constraints imposed by privacy and financial regulation. Using thematic analysis, the review identifies three recurring patterns: (i) biometric and behavioural authentication, together with anomaly detection, consistently improves detection capability relative to rule-based approaches; (ii) performance gains are offset by persistent weaknesses in explainability, demographic bias, and cross-jurisdictional data governance; and (iii) regulatory fragmentation intensifies operational uncertainty, particularly for institutions operating across regions. The synthesis further shows a continuing tension between rapid innovation in fraud prevention and the slower pace of assurance practices required for fairness, transparency, and auditability.
Bello et al. (Mon,) studied this question.