AI-Driven Threat Intelligence on Blockchain Using Deep Learning for Decentralized Cyber Risk Prediction

The increasing complexity of cyber threats such as advanced persistent threats (APTs), ransomware, distributed denial-of-service (DDoS), and smart contract exploits requires cybersecurity solutions that go beyond traditional centralized defenses. This paper proposes an AI-driven threat intelligence framework integrated with blockchain technology for decentralized and trustworthy cyber risk prediction. The novelty of the proposed framework lies in its hybrid architecture, where deep learning–based anomaly detection models (including LSTM and autoencoder networks) analyze real-time cybersecurity data—such as blockchain transaction logs, network activity records, and external threat intelligence feeds—while blockchain is used to securely store, validate, and share AI-generated threat intelligence in a tamper-resistant and decentralized manner. Unlike AI-only solutions that suffer from data integrity and trust issues, or blockchain-only approaches that lack intelligent threat detection, the proposed framework combines the strengths of both technologies to enhance detection accuracy and stakeholder trust. Experimental evaluation conducted in a simulated blockchain environment demonstrates a detection accuracy of 96.4%, a false positive rate of 3.6%, and effective identification of multiple attack categories, including smart contract exploits and 51% attacks. While the framework improves security and transparency for inter-organizational security teams, enterprise networks, and supply-chain partners, it also introduces challenges related to computational overhead and blockchain scalability. Overall, the results indicate that integrating AI-driven threat intelligence with blockchain offers a practical and robust solution for decentralized cybersecurity risk prediction.