Artificial Intelligence Applications in Fraud Risk Management: A Systematic Literature Review

Abstract

Purpose: This study aims to identify the roles, common techniques, and industry applications of Artificial Intelligence (AI) in Fraud Risk Management (FRM).

Methodology/approach: A Systematic Literature Review (SLR) was conducted following the PRISMA protocol. Data were collected from the Scopus database and managed using Zotero and SciSpace, while Microsoft Excel was used for data processing and analysis. The review covered 95 studies published between 2015 and 2025.

Results/findings: The findings show that most studies focus on fraud detection (80%), followed by risk assessment and prevention. Machine Learning (51%) and Deep Learning (29%) are the most commonly used AI techniques. The banking sector (39%) leads in implementation, followed by the insurance and public sectors. Explainable AI (XAI) is identified as an emerging trend for enhancing transparency.

Conclusions: AI greatly improves FRM by increasing detection accuracy, supporting proactive fraud prediction, and strengthening governance transparency. It transforms traditional systems into adaptive, data-driven frameworks.

Limitations: The study only used the Scopus database and found high variation among study metrics, preventing a meta-analysis.

Contribution: This study provides a comprehensive overview for researchers, practitioners, and policymakers in finance and IT. It maps the AI FRM landscape and emphasizes the importance of balancing technological performance with ethical and transparent governance.