The rapid digitalization of financial systems has led to an exponential rise in electronic transactions—and a correspondingescalation in sophisticated online fraud. Traditional machine-learning models, while effective at identifying anomalous behavior,suffer from low interpretability, mutable evidence logs, and difficulty in maintaining integrity during dispute investigations. Thispaper proposes an AI-led blockchain-based framework designed to achieve high-recall fraud detection, explainable decisionreasoning, and tamper-proof evidence management. The architecture integrates a cost-sensitive ensemble model for rare-eventdetection, an explainable AI layer for transparent reasoning, and a permissioned blockchain ledger for immutable evidenceanchoring and dispute-resolution tracking. Performance targets emphasize sub-second real-time processing and verifiable audittrails that meet regulatory expectations for accountability and transparency. The proposed framework aims to advance paymentintegrity assurance by combining the predictive power of artificial intelligence with the trust guarantees of blockchain technology.Keywords: Artificial Intelligence, Blockchain, Fraud Detection, Explainable AI, Financial Transactions, Payment Integrity,Tamper-Proof Evidence, Real-Time Analytics