AI-Driven Fraud Detection in Digital Asset Accounting: An Integrated Framework for Transaction Monitoring, Anomaly Detection, and Financial Reporting Integrity

Abstract

The rapid expansion of digital assets has created new challenges for accounting systems, particularly in detecting fraudulent transactions, maintaining reliable records, and ensuring the integrity of financial reporting. Conventional rule-based controls may be insufficient for identifying complex, high-volume, and rapidly evolving patterns of digital-asset fraud. This study examines the application of artificial intelligence (AI) to fraud detection in digital asset accounting and proposes an integrated framework linking blockchain transaction analysis with accounting controls, anomaly detection, explainable AI, and human oversight. The study develops a research model incorporating transaction characteristics, wallet behavior, temporal patterns, network relationships, and accounting discrepancies as potential fraud indicators. Machine learning and deep-learning techniques, including Random Forest, XGBoost, LSTM, and autoencoder-based approaches, are considered for identifying suspicious transaction patterns and generating fraud-risk scores. The framework further incorporates explainability mechanisms to support interpretation of AI-generated alerts and subsequent accounting verification. The study contributes an accounting-oriented perspective to digital-asset fraud detection by connecting technological identification of anomalies with reconciliation, audit procedures, professional judgment, and financial reporting controls. The proposed framework provides a structured basis for developing more responsive and auditable fraud-detection systems while recognizing the importance of data quality, model governance, and human decision-making.