Jul 2026· South Asian Journal of Business and Management Cases· Vol 15, pp. 215 - 244· 0 citations· 54 references
TL;DR
This study offers an evidence-based account of how AI in fraud detection has evolved and proposes a future research agenda emphasizing transparency, ethical assurance and global governance alignment, advancing financial risk management through conceptual clarity, methodological guidance and actionable pathways for responsible AI adoption.
Abstract
As financial fraud becomes more sophisticated and financial services are increasingly digitized, artificial intelligence (AI) and machine learning are emerging as pivotal technologies for risk management and compliance. While research into AI-driven fraud detection is advancing rapidly, the intellectual structure and theoretical underpinnings remain fragmented. This paper provides a systematic review of 118 peer-reviewed articles published between 2015 and 2025, combining bibliometric science mapping with the SPAR-4-SLR protocol to ensure rigour, transparency and replicability. Through co-word network analysis, thematic mapping and conceptual clustering, the study traces the field’s evolution from rule-based systems to adaptive anomaly detection, explainable AI and compliance models, with a focus on digital payment ecosystems and blockchain-enabled applications. The analysis highlights key theoretical anchors, including Fraud Triangle Theory, Agency Theory, Game Theory, Trust and Signalling Theories and regulatory compliance perspectives. It also identifies underexplored areas such as federated learning, algorithmic auditing and cross-jurisdictional intelligence. By mapping theoretical foundations and thematic development, this study offers an evidence-based account of how AI in fraud detection has evolved. It concludes by proposing a future research agenda emphasizing transparency, ethical assurance and global governance alignment, advancing financial risk management through conceptual clarity, methodological guidance and actionable pathways for responsible AI adoption.
This study contributes a trust-centered, infrastructure-aware AI adoption pathway specifically designed for emerging economies, offering policymakers, fintech developers, and financial institutions a pragmatic roadmap for responsible AI-enabled fraud management in Nepal.
Y. Pant, Aditya Pudasaini, R. Shrestha et al.· Islington Journal of Multidi...· 0 citations
Graph-based learning and explainable artificial intelligence (XAI) are increasingly used to improve both predictive performance and transparency in financial risk modelling. This paper presents a systematic literature review of AI and machine learning approaches for credit risk assessment and fraud detection, with specific attention to graph-based methods and explainable frameworks. Following a PRISMA-guided methodology, 149 studies published between 2015 and 2025 were analysed across multiple academic databases. The review identifies three key findings. First, graph-based models, particularly graph neural networks, can improve the modelling of relational dependencies in financial data, although their use remains more developed in fraud detection than in credit risk assessment. Second, XAI techniques such as SHAP, LIME, and rule-based methods are increasingly used to support interpretability, auditability, and regulatory compliance, but their integration with graph-based models remains limited. Third, recent research is shifting from purely predictive modelling towards deployment-oriented financial AI systems that address class imbalance, concept drift, scalability, real-time monitoring, and governance. To address these gaps, this paper proposes a deployment-oriented reference framework that links graph construction, graph-based modelling, explainability mechanisms, monitoring, human oversight, and governance controls. The findings provide a structured synthesis of current research and practical guidance for developing scalable, transparent, and accountable financial AI systems.
Zahra Shams Khoozani, M. Seera, C. Lim· Artificial Intelligence Revi...· 1 citation
A conceptual model demonstrating how AI-enabled analytics techniques -- encompassing supervised machine learning, unsupervised anomaly detection, deep learning, and graph-based network analytics -- directly and indirectly enhance fraud detection accuracy, response speed, and organisational risk posture is developed.
Prof. Roopa U Prof. Roopa U, Shrushti S Nelogi Shrushti S Nelogi· International Scientific Jou...· 0 citations
The study introduces the Stability–Transparency–Reliability (STR) model, which advances TPB, Socio-Technical Systems Theory, and the Dynamic Capabilities View by reframing XAI from a static interpretability output into a recursive governance capability, formalised through the concept of Interpretative Agility, with direct implications for financial institutions operating under the EU AI Act.
Background: Artificial Intelligence (AI) has revolutionized tax compliance, improving fraud detection, risk assessment, and voluntary compliance. The knowledge landscape, intellectual bases and the trends in this nearing field, however, are not sufficiently well articulated and require a systematic mapping.
Methods: The structured search strategy was applied to the PRISMA methodology to retrieve literature on AI and tax compliance from Scopus. The number of records analyzed is 527 records published between 2004 and 2026, and it was done through Biblioshiny and VOSviewer. The publication trends, prolific authors, journals, countries, and citations, were analyzed to explore the performance, and science mapping included co-citation, bibliographic coupling, keyword co-occurrence, and thematic evolution.
Results: Research has been growing at an exponential rate since 2020, thanks to developments in machine learning, big data analytics and digital tax administration. Leading contributors became China, Germany and the United States. Focusing on the dominant themes in the field of taxation, artificial intelligence, machine learning, fraud detection and data mining. The growing research shows the importance of explainable AI, governance, transparency, public trust and ethical AI adoption.
Conclusion: The research suggests a five-step framework for implementing AI for tax compliance: risk concept formulation, data integration, model selection, governance and explainability, and continuous monitoring.
Novelty: This review article not only offers a comprehensive bibliometric mapping of AI and tax compliance, but also combines performance mapping with science mapping, and selects emerging themes and new research directions.
Dinesh Bidari, Nasiruddin Molla· NPRC Journal of Multidiscipl...· 0 citations
Digital transformation has sharply increased fraud risks in banking, with global annual losses exceeding $50 billion. This systematic review analyzes 20 influential peer-reviewed studies from 2024 to 2025, selected through a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020–guided search from over 150 papers in IEEE Xplore Digital Library, SpringerLink, ScienceDirect, Scopus, and Google Scholar, using terms such as "AI banking fraud detection," "deep learning financial fraud," "graph neural network fraud," and "explainable AI banking." Included studies are published in reputable venues, report robust evaluations on real or large-scale transaction data, introduce meaningful innovations (e.g., novel architectures or privacy-preserving methods), and address deployment in banking environments. The reviewed works employ deep neural networks, graph-based models, hybrid ensembles, explainable artificial intelligence (XAI), and blockchain-inspired components, achieving detection accuracies of 93–96% on real transaction datasets. Nonetheless, major obstacles still limit large-scale adoption. We identify 10 recurring challenges; high computational costs, data quality problems, and limited model interpretability each appear in at least 30% of the studies. Our analysis compares the strengths, weaknesses, and contributions of existing approaches and highlights critical gaps in scalability, robustness, and standardization. We recommend wider use of lightweight models, broader application of federated learning, creation of shared benchmark datasets, and adoption of common evaluation protocols. Finally, we outline priority directions to move beyond a narrow focus on accuracy toward fraud detection solutions that are scalable, secure, transparent, and cost-effective, enabling more confident deployment of AI-based systems in real-world banking.
Received: 14 November 2025 | Revised: 14 February 2026 | Accepted: 14 July 2026
Conflicts of Interest
The authors declare that they have no conflicts of interest to this work.
Data Availability Statement
Data sharing is not applicable to this article as no new data were created or analyzed in this study.
Author Contribution Statement
Hamid Banirostam: Conceptualization, Methodology, Formal analysis, Investigation, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration. Elham Shamsinejad: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Data curation, Writing – original draft, Writing – review & editing.
Hamid Banirostam, E. Shamsinejad· FinTech and Sustainable Inno...· 0 citations