CBAATM: A Blockchain-AI Integrated Framework for Real-Time Anomaly Detection and Compliance Verification in Smart Accounting Information Systems

Abstract

Accounting is undergoing a radical transformation due to the integration of traditional information systems with blockchain technology and artificial intelligence. Openness, automation, and smart decision-making will all become a reality via this connection. However, traditional SAIS are typically centralized and do not inherently include blockchain or AI. In this study, Smart Accounting Information System (SAIS) technologies are redefined through the integration of these technologies to enhance transparency, automation, and real-time assurance. Blockchain technology's immutability, traceability, and AI's ability to recognize abnormalities and predict provide a more intelligent and secure auditing process. Conventional accounting methods have several issues, including delayed audits, lack of transparency, fraud, and human mistakes. Existing systems fail to provide intelligent anomaly detection and real-time transaction traceability. Financial reporting and audits need immutable records and proactive analytics. There is an urgent need for a single framework to ensure this requirement and its quick implementation. This study proposes the collaborative blockchain-AI audit trails method (CBAATM) for Smart Accounting Information Systems. This is done due to the difficulties mentioned. AI-powered modules utilize fuzzy inference to dynamically analyze audit risks and Random Forest classifiers to detect real-time fraud. This research project utilizes zero-knowledge proofs and homomorphic encryption to simultaneously handle data aggregation, privacy, and independent audits. Using middleware application programming interfaces makes integration with ERP and AIS systems easy. Throughout the testing process, the model outperforms conventional audits. The methodology, according to statistical research, ensures the detection accuracy ratio of 95%, integrity of the blockchain 99.2% of the time, identifies abnormalities 94.1% of the time, satisfies compliance standards 95.4% of the time, and reduces audit latency by 41.5% compared to other existing models.