GOVERNANCE-AWARE PHISHGUARD++: A ZKP-DID AND GRAPH NEURAL NETWORK FRAMEWORK FOR ACCOUNTABLE CYBERSECURITY IN LOCAL DIGITAL PUBLIC SERVICES

Povzetek

Digital technologies are increasingly used in the delivery of local public services, which makes it important to have strong authentication, privacy, and accountable access control on digital governance platforms. In this study, an authentication solution named Governance-Aware PhishGuard++ from the Web3 space is proposed that combines Zero Knowledge Proofs (ZKPs), Decentralized Identifiers (DIDs), and Graph Neural Network (GNN) based threat intelligence to augment the identity verification process, safeguard sensitive information, and enhance the capability of phishing threat detection. The principal contribution is called a Governance-Aware Risk-Adaptive Access Control (GRAAC) mechanism, which combines identity verification, dynamic phishing-risk estimation, institutional authorisation policies, and accountable access enforcement. The proposed extension differs from the original framework in that it also considers factors such as the probability of a threat, institutional permission, resource sensitivity, and governance requirements in its evaluation of access requests rather than just the mutual authentication and real-time threat detection in the original framework. Only a policy-constrained decision mechanism will decide whether to allow or disallow requests as well as whether to escalate to authorised review. Enforcement and privacy-conscious audit records via smart-contracts or the application level support traceability without necessarily revealing sensitive identity information. It will be tested using the available phishing data sources and synthetic municipal-services phishing scenarios. The experiments will compare accuracy of phishing detection, unauthorised access acceptance, denial of legitimate access requests; policy compliance, latency in decision making and computation overhead; and completeness of auditing. The proposed mechanism will be evaluated by means of baseline comparison, ablation studies, and sensitivity analysis. The research aims to combine explicit institutional governance policies with privacy-preserving authentication and graph-based threat intelligence to lay a foundation for measurable and accountable cybersecurity for local digital public services.