ScanCanvas: A Review of AI-Driven Web Vulnerability Scanners with LLM-Based Enhancements

This review paper explores advancements in web vulnerability scanning through the integration of Artificial Intelligence and Large Language Models (LLMs). It analyzes existing scanners such as OWASP ZAP, Nikto, and SQLMap, highlighting their limitations in contextual reasoning and prioritization. The paper surveys frameworks like YURASCANNER, AutoPentest, Rogue, and Burp AI to demonstrate how LLMs enhance automation, reasoning, and adaptive scanning. Future prospects include intelligent threat modeling, multimodal analysis, and private LLM deployment. The work also reflects on the practical case study of ScanCanvas — an LLM-powered vulnerability detection framework.