AIDefender: An AI-Powered Proactive Auditing and Threat Mitigation Framework for Decentralized Storage Networks

Abstract

Decentralized Storage Networks (DSNs) face a critical security limitation: their reliance on periodic cryptographic proofs (PoRep/PoSt) creates a reactive model with inherent “detection lag,” leaving them vulnerable to sophisticated, slow-burn attacks, such as targeted data withholding and gradual corruption. To address this issue, we propose AIDefender, a novel AI-powered framework that shifts the DSN security from reactive attestation to proactive threat mitigation. AIDefender employs a two-tiered architecture: 1) Lightweight On-Node Auditors (LONA) perform continuous, verifiable anomaly detection on privacy-preserving node telemetry, and 2) a Federated Threat Intelligence Hub (FTHI) enables collaborative, privacy-conscious model evolution across the network. Our framework directly integrates with blockchain smart contracts to enable automated mitigation responses, such as proactive data redundancy. Evaluation on a simulated 200-node testbed demonstrated that AIDefender predicted 94% of attacks with a mean lead time of 14.3 h, preventing 76-95% of potential data loss while introducing less than 5% computational overhead. This represents a fundamental paradigm shift towards intelligent, resilient, and proactive security for decentralized storage ecosystems.