An intelligent model based on federated learning and explainable AI for combating digital fraud and corruption in banks: An applied comparative study

Purpose and Design: This research aims to develop and empirically validate an intelligent digital audit model leveraging Federated Learning (FL) and Explainable Artificial Intelligence (XAI) to prevent and detect digital fraud and corruption within banks listed on the Egyptian Stock Exchange (EGX). The study responds to urgent calls for technological solutions to mitigate risks arising from digital fraud perpetrated by both external actors and internal stakeholders, including auditors and governance officers. It also seeks to support regulatory reform efforts by the Egyptian Financial Regulatory Authority (FRA). Method and Approach: Employing a mixed-methods research design, the study integrates quantitative survey analysis and qualitative elite interviews. A structured survey was administered to 120 audit professionals, governance officers, and financial executives in EGX-listed banks. Additionally, 15 semi-structured interviews were conducted with international audit and regulatory experts from developed markets such as the UK and US. Partial Least Squares Structural Equation Modeling (PLS-SEM) tested the research hypotheses, while thematic analysis explored cross-market insights. A comparative case study analysis was also performed to benchmark Egypt’s governance practices against established regulatory models abroad. Findings: Results confirm that integrating FL and XAI significantly enhances audit quality, strengthens internal control evaluation, and improves fraud and corruption detection in EGX-listed banks. Governance quality mediates these benefits, while digital maturity moderates the successful deployment of FL-XAI systems. Comparative insights reveal stark differences between Egypt and developed markets in regulatory readiness, technological adoption, and auditor accountability, highlighting critical areas for policy intervention. Originality and Value: This study represents the first empirical application of an integrated FL-XAI audit framework focused on preventing and detecting digital fraud and corruption in an emerging market banking sector. It uniquely bridges technical innovation with practical regulatory implications, offering a novel pathway for legislative reforms by the FRA. Theoretical, Practical, and Social Implications: Theoretically, the research extends Technology Acceptance Model (TAM), Agency Theory, and Governance Theory in digital audit contexts. Practically, it offers actionable recommendations for Egyptian regulators and audit professionals, including potential legal mandates for digital audit tools. Socially, it contributes to financial integrity, investor protection, and sustainable governance, aligning with Egypt Vision 2030.