A privacy protection system for personal sensitive information using local ai redaction and verifiable credential (VC) access control

Abstract

The present invention is related to a privacy protection system for personal sensitive information using local AI Redaction and Verifiable Credential (VC) Access Control. This invention relates to a privacy-aware camera system capable of performing on-device AI-powered detection and redaction of sensitive information such as human faces, textual data, and vehicle license plates in real time. The system operates entirely at the edge, ensuring that original unprocessed content never leaves the local device without authorization. A mock Verifiable Credential (VC) and Decentralized Identifier (DID) framework is integrated to authenticate and grant controlled access for selective unblurring of redacted regions, enabling context-based privacy enforcement. The architecture consists of five layers: input & capture, AI detection & redaction, access control, selective cloud inference, and logging & compliance. It supports rule-based privacy modes, dynamic configuration, and maintains a complete audit trail through secure logging for regulatory compliance. The system is suitable for privacy-critical applications such as surveillance, document digitization, vehicle monitoring, and secure archival, where strict control over sensitive visual data is required.