This thesis addresses the tension between financial inclusion and regulatory compliance. More than 1.4 billion adults remain outside the formal system, while many more are underserved, especially low-income undocumented populations. At the same time, illicit finance requires Know Your Customer (KYC) and Anti-Money-Laundering (AML) controls. Regulators must protect financial integrity without reinforcing exclusion.
The Financial Action Task Force (FATF) supports risk-based approaches allowing simplified KYC/AML for low-risk clients and products. However, inclusion decisions are also shaped by compliance costs, liability, and institutional incentives. Many underserved users therefore remain unattractive to serve, even when they are low risk.
Decentralized Finance (DeFi) and Distributed Ledger Technologies (DLTs) create new possibilities for inclusion. By enabling programmable peer-to-peer interactions, DeFi can use decentralized social trust and cryptographic proofs to reduce costs and include users whom traditional institutions might exclude. Yet permissionless systems raise challenges around accountability, identity, privacy, and compliance.
The thesis introduces a Computational Trust and Risk Engine (CTRE) for risk-based compliance and inclusion. The CTRE combines recognition schemes, privacy-enhancing technologies, rule-based reasoning, probabilistic assessment, and adaptive trust computation. It enables proportional, explainable, and verifiable KYC/AML decisions using trust and risk indicators from on-chain and off-chain sources.
The first contribution is an accretionary proof-of-address scheme based on decentralized social trust. Designed for poverty and informality contexts, it supports alternative KYC methods aligned with FATF recommendations. It combines social trust, cryptography, and minimal-data certificates. A field pilot in Rocinha, Brazil, showed feasibility and acceptance, demonstrating community recognition as portable, privacy-preserving proof.
The second contribution is a formal rule language translating regulatory requirements into structured logic. Based on KYC/AML requirements across jurisdictions, it reduces ambiguity, expresses trust and risk formulas, and supports dynamic decision-making. In a peer-to-peer marketplace, participants assessed trustworthiness and regulatory conformity without intermediaries while preserving compliance evidence.
The third contribution develops dynamic computational models combining rule-based compliance reasoning with adaptive trust mechanisms. In simulations, participants approve transactions based on evolving trust and risk indicators. Results show that time-based trust growth and dispositional trust in verifiers increase approval rates and monetary value while remaining consistent with risk-based KYC/AML principles. Dynamic reasoning can complement fixed-rule compliance models and include excluded users.
