A Comprehensive Result Paper on UnveilThreatAI: See the Invisible Threat in What You Share

Abstract

The project, “UnveilThreatAI: See the Invisible Threats in What You Share,” provides an AI-powered system to support users in identifying cybersecurity risks hidden in images, documents, and URLs. The platform combines image metadata inspection, document parsing, suspicious link analysis, explainable AI, and risk visualization to help users make safer sharing decisions. The system is implemented as a web-based application using React for the frontend, Django REST Framework for the backend, PostgreSQL for storage, and Python-based AI/ML utilities for analysis. It generates a consolidated report that includes a severity score, metadata exposure findings, suspicious URL indicators, and a consequence chain showing how detected issues may lead to phishing, privacy loss, identity theft, or financial fraud. Experimental observations show that the system processes moderate-size inputs reliably and achieves observed detection performance in the range of 80–90% on representative test samples. The platform improves user awareness by combining automated detection with simple, interpretable output.