A Blockchain Solution for Decentralized Content Verification and its Application to Deepfake Detection and Fintech Credit Scoring
Abstract
本文提出了一个基于区块链的去中心化内容验证框架，具有两个应用领域：检测人工智能生成的深度伪造品和金融科技领域的自动化信用评分。这项工作引入了一种混合的链上/链下架构，其中独立证明者对链外内容进行评分，最终结果锚定在链上以实现公共审计，形成 Attestia 协议的科学基础。
重点议题
主要贡献
Hybrid Architecture
A novel on-chain/off-chain design where attesters score content independently off-chain, and final outcomes are anchored on-chain via the Ethereum Attestation Service (EAS) for public auditability.
Deepfake Detection Application
The framework is applied to AI-generated media detection, demonstrating how decentralized consensus can produce more reliable and tamper-resistant authenticity scores than centralized models.
Fintech Credit Scoring
A second application domain explores how the same decentralized verification infrastructure can automate and audit credit scoring processes in the financial sector, improving transparency and reducing bias.
Incentive & Slashing Mechanisms
A staking-based incentive model is formally described: attesters earn rewards aligned with consensus and face stake slashing for deviating scores, ensuring honest participation at protocol level.