Bridging the Gap Between Research and Practice in Intelligently Bypassing WAF
summary
Claims an automated bypass pipeline that uses PPO reinforcement learning to sequence small, expert-derived atomic mutations and a grammar- and lexical-based mutation step to preserve payload semantics. Training splits vendor sandboxes and attack targets to cut attempts; experiments generate ~20,000 executable samples, extract ~20 black‑box bypass modes, find >50 white‑box modes, and achieve nearly 100% success against multiple commercial WAFs. Solution is integrated into vendor rules and verified as a “one‑click injection” against dozens of customers.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 0:57Agenda
- 1:53Related Works
- 7:15Key Issues
- 9:15Our Research
- 13:28Our Workflow
- 14:20Our Scheme
- 16:33Device Environment
- 18:07Semantic Environment
- 19:36Core Algorithm
- 23:33Experimental Results
related talks
37mWeb Application Firewall Bypassing
35mAutoSpear: Towards Automatically Bypassing and Inspecting Web Application Firewalls
43mWAF Bypass Techniques Using HTTP Standard and Web Servers’ Behavior
45mAutomatic Discovery of Evasion Vulnerabilities Using Targeted Protocol Fuzzing
43mMaking WAF Mainstream: From Static Defenses to... (Roy Weisfeld, Surya Pentakota)
46mIt is rough without a WAF
40mWAF Made Easy
47mReversing Engineering a Web Application - For Fun, Behavior & WAF Detection