Dr. Mohit Pant, Rashmi Pant

DOI: https://doi.org/10.5281/zenodo.22356265

The rapid adoption of Mobile Money Payment Systems (MMPS) has increased financial inclusion, particularly in developing countries such as India, China, Brazil and several African countries. But the increasing use of digital payment platforms has also resulted in the increasing of sophisticated fraud attacks such as SMS phishing, account impersonation, transaction manipulation, and social engineering. The existing fraud detection techniques are largely based on static machine learning and deep learning models that are not able to adapt to changing fraud patterns, have limited interpretability and do not utilize unstructured text. In this paper, we have developed an Adaptive Explainable Multi-Agent Fraud Detection Framework (AEMAF) which combines LLMs, hybrid machine learning and Explainable Artificial Intelligence (XAI) for the fraud detection of MMPS. The framework uses smart agents in SMS semantic analysis, transaction risk assessment, behavioral anomaly detection, device profiling, decision fusion and explanation generation to identify and assess fraud risk and their corresponding behaviors. A multimodal feature fusion module combines these heterogeneous data sources to improve detection accuracy. A feedback system based on an adaptive algorithm continuously updates the detection model on verified fraud cases to protect from concept drift. SHapley Additive exPlanations (SHAP) based on LLM natural language explanations allows the fraud detection system to be transparent and interpretable. The proposed framework will improve fraud detection accuracy, reduce false positives, improve model adaptability and be more secure and transparent for mobile financial services.

Mobile Money Payment Systems, Fraud Detection, Large Language Models, Explainable Artificial Intelligence, Multi-Agent Systems, Hybrid Machine Learning, SHAP, Adaptive Learning.

This paper has presented an adaptive explainable multi-agent fraud detection framework (AEMAF) to improve fraud detection in Mobile Money payment systems. Its framework combines LLMs, hybrid machine learning, explainable AI, and adaptive learning to provide an efficient and robust framework for analyzing structured transaction data and unstructured SMS data in a unified framework. Through the application of specialized intelligent agents for transaction analysis, behavioral profiling, device monitoring, semantic SMS understanding, feature fusion, and decision making, the framework provides a holistic solution to fraud detection. The hybrid machine learning engine combines multimodal features to enhance fraud classification, and the SHAP explainability module is transparent in terms of which factors play a role in each prediction. Moreover, the adaptive learning model is able to update the model and its system to verify fraud cases and is robust against concept drift as well as evolving cyber attacks. In comparison, the proposed AEMAF framework achieved superior performance to conventional machine learning and transformer-based approaches in terms of accuracy, precision, recall, F1-score, and ROC-AUC. Multimodal learning, explainable AI, and adaptive feedback support the detection capability and provide interpretable decision support for financial institutions. In conclusion, the proposed framework is a scalable, intelligent, and transparent solution for next-generation mobile payment security

[1] GSMA, State of the Industry Report on Mobile Money 2024, GSMA, London, UK, 2024.
[2] E. A. Lopez-Rojas, A. Elmir, and S. Axelsson, “PaySim: A financial mobile money simulator for fraud detection,” in Proc. 28th European Modeling and Simulation Symposium (EMSS), Larnaca, Cyprus, 2016, pp. 249–255.
[3] Kendall, J., Maurer, B., Machoka, P., & Veniard, C. (2011). An emerging platform: From money transfer system to mobile money ecosystem. Innovations: Technology, Governance, Globalization, 6(4), 49-64.
[4] Bezovski, Z. (2016). The future of the mobile payment as electronic payment system. European Journal of Business and Management, 8(8), 127-132.
[5] Sayibu, U., Asante, M., Abdul-Salam, G., & Frimpong, T. (2025). Fraud Prediction and Prevention in Mobile Money Payment Systems (MMPS): A Systematic Literature Review of Text‐Based Detection Methods. Security and Communication Networks, 2025(1), 8913715.
[6] Lokanan, M. E. (2023). Predicting mobile money transaction fraud using machine learning algorithms. Applied AI Letters, 4(2), e85.
[7] Raghavan, P., & El Gayar, N. (2019, December). Fraud detection using machine learning and deep learning. In 2019 international conference on computational intelligence and knowledge economy (ICCIKE) (pp. 334-339). IEEE.
[8] Hashemi, S. K., Mirtaheri, S. L., & Greco, S. (2022). Fraud detection in banking data by machine learning techniques. IEEE Access, 11, 3034-3043.
[9] Almazroi, A. A., & Ayub, N. (2023). Online payment fraud detection model using machine learning techniques. Ieee Access, 11, 137188-137203.
[10] Ali, A. A., Khedr, A. M., El-Bannany, M., & Kanakkayil, S. (2023). A powerful predicting model for financial statement fraud based on optimized XGBoost ensemble learning technique. Applied Sciences, 13(4), 2272.
[11] Ileberi, E., Sun, Y., & Wang, Z. (2021). Performance evaluation of machine learning methods for credit card fraud detection using SMOTE and AdaBoost. IEEE access, 9, 165286-165294.
[12] Oliveira, D. N. O. (2026). Neural networks for real-time financial fraud detection. Journal International Review of Research Studies, 1(03), 1-49.
[13] Feng, P. (2025). Hybrid BiLSTM-transformer model for identifying fraudulent transactions in financial systems. Journal of Computer Science and Software Applications, 5(3).
[14] Zafar, U., & Wu, F. (2026). Methodological challenges in explainable AI for fraud detection: a systematic literature review. Artificial Intelligence Review, 59(4), 115.
[15] Yan, Y., Hu, T., & Zhu, W. (2024, December). Leveraging large language models for enhancing financial compliance: A focus on anti-money laundering applications. In 2024 4th international conference on robotics, automation and artificial intelligence (RAAI) (pp. 260-273). IEEE.
[16] Erva Ergun, Z., & Sefer, E. (2025). Financial statement fraud detection via large language models. Intelligent Systems in Accounting, Finance and Management, 32(4), e70021.
[17] Wu, Y., & Li, Y. (2026). Agentic FinTech: A Comprehensive Survey on AI Agents in Finance in the Era of LLMs. Available at SSRN 6136529.
[18] Ward, E., & Iqbal, Z. (2026). The Future of FinRisk: Integrating Multi-Agent LLMs and Machine Learning for Agentic Fraud Detection.
[19] Pant, M., & Malviya, L. (2023). Design, developments, and applications of 5G antennas: a review. International journal of microwave and wireless technologies, 15(1), 156-182.
[20] Pant, M., & Malviya, L. (2024). SIW MIMO antenna with high gain and isolation for fifth generation wireless communication systems. Frequenz, 78(9-10), 479-497.
[21] Oliveira, D. N. O. (2026). Neural networks for real-time financial fraud detection. Journal International Review of Research Studies, 1(03), 1-49.

Dr. Mohit Pant, Rashmi Pant (2026). Adaptive Explainable Multi-Agent Fraud Detection Framework Using Large Language Models and Hybrid Machine Learning for Mobile Money Payment Systems. International Journal of Computer Science Engineering Techniques, 10(5), 145–154. ISSN: 2455-135X. DOI: https://doi.org/10.5281/zenodo.22356265