Secure Data Sharing for AI Training Using Blockchain Ledger

The increasing demand for Artificial Intelligence (AI) model training has intensified the need for large-scale, high-quality data. However, concerns over data security, ownership, and privacy continue to limit the sharing of valuable data across institutions. To address this issue, the Access Control and Data Source Tracking (AC-DPT) proposed system introduces a blockchain-based framework for secure, transparent, and incentivized data sharing for AI training. The AC-DPT model enhances user authentication by using a blockchain-based approach with smart contracts to record and verify transactions. Furthermore, the provide a Data Anonymization with Federated Encryption (DAFE) method to protect data privacy through optimization and public-key encryption before data is shared. Then, implemented a Decentralized Data Market (DDM) method to obtain data based on user quality, develop a reward mechanism and collaboration. Finally, demonstrates improved accuracy through simulation results by improving contributor engagement through dynamic incentive distribution using the proposed AC-DPT method. Furthermore, the system achieves 93 % accuracy in validating access requests through simulation parameters, including data access latency, throughput, model accuracy, and tamper detection.