Impact of Artificial Intelligence on Fraud Detection in Retail Banking: A Quantitative Study

Abstract

Background: Fraud in retail banking has taken a dynamic and technologically-oriented threat environment. Conventional rule-based fraud detection solutions are no longer adequate in detecting the new fraud schemes and sealing the operational loopholes. The AI-based fraud detection systems provide adaptive anomaly detection, behavioral modeling, and risk scoring in real-time. Nevertheless, predictive capability will not be effective alone in such systems, but also in the quality of system design and organizational enablement in frontline banking operations. Purpose: This paper will analyze the role of AI System Design Quality and AI Organizational Enablement in the effectiveness of Fraud Control in retail banking. It also explores how Perceived Loophole Detection Capability (PLDC) plays a mediating role in the transfer of AI system attributes to better fraud containment results. Methods: The quantitative cross-sectional survey design was used among the retail banking operations and call-center employees, who dealt with fraud alerts. Analysis of data was performed with the help of Partial Least Squares Structural Equation Modeling (PLS-SEM) according to the recommended methodology (Hair et al., 2022). Constructs were conceptualized into reflective-reflective higher-order elements. Results: The results show that AI System Design Quality and Organizational Enablement have a positive and significant impact on PLDC that consequently improves Fraud Control Effectiveness. The mediation analysis proves that the relationship between AI system factors and the results of the fraud control is mediated by PLDC partially. Conclusion: AI is not only used to increase technical accuracy but also perceived ability to detect operational blind spots, which are used to mitigate fraud. To maximize the performance of fraud control, it is important to have effective system design and organizational support.