Current blockchain protocols (e.g., Proof-of-Work and Proof-of-Stake) secure the ledger yet cannot measure validator trustworthiness, allowing subtle misconduct that is especially damaging in decentralized-finance (DeFi) settings. We introduce Proof-of-Behavior (PoB), a consensus model that (i) gives each action a layered utility score -- covering motivation and outcome, (ii) adapts validator weights using recent scores, and (iii) applies decentralized verification with proportional slashing. The reward design is incentive-compatible, yielding a Nash equilibrium in which honest behavior maximizes long-run pay-offs. Simulated DeFi experiments (loan-fraud detection, reputation-weighted validation) show that PoB cuts fraud acceptance by more than 90%, demotes malicious validators within two rounds, and improves proposer fairness versus standard PoS, all with no more than a 5% throughput overhead. By linking consensus influence to verifiably trustworthy conduct, PoB offers a scalable, regulation-friendly foundation for secure and fair blockchain governance in financial applications.