🤖 AI Summary
This study addresses the vulnerability of high-accuracy fraud detection models in enterprise settings, where insiders may tamper with logs or bypass approval workflows, thereby compromising audit integrity. To counter this, the authors propose a tamper-proof AI auditing system that uniquely leverages blockchain as an active enforcement layer. By anchoring explainable machine learning predictions and approval workflows atomically to a Layer-2 chain (e.g., Polygon) via smart contracts, the system ensures end-to-end immutability, verifiability, and compliance with GDPR. Empirical evaluation demonstrates strong performance: an F1 score of 0.895, a PR-AUC of 0.974, inference latency under 25 milliseconds, transaction costs below \$0.01 per operation, and scalability to support tens of thousands of monthly payments.
📝 Abstract
In enterprise fraud detection, model accuracy alone is insufficient when insiders can tamper with audit logs or bypass approval workflows. Real-world incidents show that fraud often persists not because detection algorithms fail, but because the audit trail itself is controllable by privileged operators. This exposes a fundamental trust gap: *who audits the auditor?*
We present a tamper-evident fraud detection system that anchors both ML predictions and workflow execution to an immutable blockchain ledger. Rather than using blockchain as passive storage, we enforce the entire approval process through smart contracts, ensuring that every transaction, prediction, and explanation is atomically recorded and cannot be retroactively modified. Our detection module achieves competitive accuracy (F1 = 0.895, PR-AUC = 0.974) while providing cryptographically verifiable decision trails that support regulatory auditability requirements (e.g., GDPR Article 22). System evaluation shows sub-25 ms inference latency and economically viable deployment on Layer-2 networks at under \$0.01 per transaction (validated against PolygonScan data), supporting enterprise-scale workloads of 10,000+ monthly payments.