A Distributed Framework for Adaptive Cybersecurity via Convergence of Parallel AI and Blockchain Architectures

Abstract

As the digital landscape expands, centralised cybersecurity frameworks grow increasingly vulnerable to sophisticated threats, creating single points of failure and targets for adversarial data manipulation. While AI enables real-time threat detection and big data analytics, its centralised deployment limits efficacy and exposes training data to poisoning and evasion attacks. To address this, the AICyber-Chain model proposes a distributed framework combining parallel AI and blockchain architectures. It leverages a hybrid Proof-of-Stake (PoS) and Byzantine Fault Tolerance (BFT) mechanism with IPFS and Private Data Centres (PDCs) for secure decentralised storage and processing. Generative Adversarial Networks (GANs) refine security rules, while Ethereum-based smart contracts enable automated responses and trustless data sharing. Results on the Rinkeby test network show 1.8× faster authentication, 25% lower gas consumption, F1 score of 0.92, and 1.2 s response time, with a medical data sharing use case ensuring data provenance and tamper-proof control.