ZK-Credit: Privacy-Preserving Credit Scoring

Abstract

The integration of Zero-Knowledge Proofs (ZKPS) with blockchain for credit scoring represents a novel approach to preserving privacy in financial transactions. This paper explores the untapped potential of ZKPS to securely verify creditworthiness without revealing sensitive financial information, addressing a significant gap in the current research landscape. The scarcity of exploration in this domain highlights the pioneering nature of this project, suggesting it could radically transform privacy preservation in financial assessments. By proposing a method that ensures financial data privacy and meets regulatory demands, this study not only challenges traditional credit evaluation methods but also sets a new standard for secure financial transactions. This research introduces an end-to-end solution that combines ZKPS, machine learning and blockchain technology to enhance privacy in credit scoring systems within banks and financial institutions. This approach facilitates the confidential verification of credit scores through generating and verifying zero knowledge proofs and persisting them on the blockchain. The novelty of this study lies in its capacity to affect a paradigm shift in financial assessments, promising a future where trust and security in the banking sector are redefined through enhanced privacy measures. Model testing results indicate an extremely high performance (99% accuracy) in the ML model responsible for generating the credit scores. The selected zero knowledge trusted setups (Grothl6) and proof sizes are optimized for the use case and provide superior performance over alternatives.