Beyond the Veil: Analyzing Dark Web Threats with AI-Based Detection and Mitigation

This paper presents "Beyond the Veil," an AI-based cybersecurity system for detecting and mitigating threats on dark web (onion) websites. The system integrates machine learning and NLP-based content classification with a specialized vulnerability scanner targeting XSS, Clickjacking, CORS misconfigurations, HTTP security headers, JavaScript exploits, and HTML Injection. The vulnerability scanner achieved 100% detection accuracy with 36.7% improved memory efficiency over baseline systems. The AI mitigation module uses the Gemini API to generate real-time remediation suggestions. Testing was conducted across 89 onion websites. This work was submitted as a final year B.E. project at G. V. Acharya Institute of Engineering & Technology, Department of Computer Science Engineering (AI & ML), guided by Dr. Dinesh Bhere.