Decentralized and Explainable Credit Scoring: A Trust-Centric Fintech Approach

The rapid expansion of digital lending demands credit-scoring systems that are not only accurate but also transparent, auditable, and resilient to bias. This paper proposes a decentralized and explainable credit-scoring framework that integrates blockchain-based trust infrastructures with interpretable machine learning to create a trust-centric fintech architecture. While traditional credit-scoring models rely on opaque, centralized data custodianship, our approach distributes data verification and model accountability across a permissioned blockchain, enhancing trust in line with cross-disciplinary perspectives on relational and institutional trust [1] and leveraging trustless ledgers as theorized in emerging blockchain adoption research [1]. Building on recent advances in trusted AI enabled by blockchain–LLM convergence [3], the proposed system embeds explainability modules that provide borrower-level rationales and regulator-auditable model pathways. By framing decentralized credit scoring as a combinative capability [4], the architecture enables multi-institution collaboration without exposing proprietary data. Additionally, the model aligns with private-ordering regulatory mechanisms [5], providing tamper-evident audit trails suitable for fintech compliance. Experimental validation on synthetic and real-world lending datasets demonstrates improvements in fairness, transparency, and stakeholder trust, establishing a viable foundation for next-generation responsible credit-scoring ecosystems.