WebMaze diverts AI-powered web scanners with interactive decoys
WebMaze: Defending Against LLM-Powered Web Scanners with Functionality-Coupled Deception
Cryptography and Security
Summary
Automatic programs powered by large language models (LLMs) can scan websites to find vulnerabilities, making it easier for attackers to target them. The authors studied how these AI scanners work and created WebMaze, a defense that keeps websites fully usable while misleading scanners away from real pages. WebMaze places hidden traps in genuine pages that lead scanners into fake, interactive decoy pages, so attackers spend most of their time exploring false content. Even when scanners are told about the trick, they still mostly explore decoys, protecting the real site.
What this means in practice
- •For web security teams: Protect web applications from automated AI scanning by diverting attackers toward interactive decoy pages without impacting legitimate users.
- •For internet infrastructure operators: Integrate deception defenses to reduce the risk of AI-driven reconnaissance affecting hosted web services at scale.
Authors
Pengfei Zeng, Changhua Luo, Penghui Li
Abstract
Large language models (LLMs) have enabled more effective vulnerability scanning of web applications. However, these capabilities also make it significantly easier for attackers to automate cyberattacks on the web. In this work, we investigate how to divert malicious LLM-based web scanning, with preservation of genuine functionality as a deployment requirement. We systematically study six LLM-based scanners and summarize how they gather information and iteratively select actions based on observations. Guided by the study, we then present WebMaze, a defense that preserves genuine application functionality while diverting LLM-based scanners to a generated decoy world. WebMaze embeds functionality-coupled traps in genuine pages and constructs view-tailored interactive decoy worlds from the application's existing views. The traps blend into genuine components and expose entry points to decoy worlds that sustain exploration through connected pages and interactions without changing genuine functionality. Across 12 web applications and three LLM-based scanners under the default no-disclosure setting, decoy pages account for an average of 93.0% of the distinct pages requested per run. This demonstrates that WebMaze effectively diverts scanners away from genuine application pages and into the generated decoy worlds. Even when explicitly informed about the decoy mechanism and instructed to avoid decoy pages, scanners continue to explore predominantly decoy pages.