Today for AI
HOT RADAR
agentHEAT 11.1°

AI CURATED HOT STORY · 10/21/2026

Cloudflare Builds LLM Black-Box Framework to Test WAF Defenses

1 reports archived1 independent sourcesupdated 10/21/2026, 11:35:05
Synthesis & Latest Updates
1 Sources Cross-Validated

Cloudflare developed a Python-based framework using LLMs to dynamically generate and iterate web attack variants for black-box WAF evaluation. Across 45 scenarios, the system adjusted payload encoding and delivery via feedback loops, yielding 49 valid findings that drove three managed rule updates, demonstrating structured AI application in security testing.

LATEST/Cloudflare developed a Python-based framework using LLMs to dynamically generate and iterate web attack variants for black-box WAF evaluation. Across 45 scenarios, the system adjusted payload encoding and delivery via feedback loops, yielding 49 valid findings that drove three managed rule updates, demonstrating structured AI application in security testing.

HEAT TRENDHourly heat curve

4 fully observed hours
Now
11.1
Peak
11.912:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

TIMELINECoverage timeline

Total 1 reports · Latest first
  1. InfoQ 中文T2·68 pts
    • Black-box approach: LLMs adjust payloads based solely on WAF response feedback without accessing internal rules or source code.
    • Human-in-the-loop architecture: A Python framework handles request orchestration and state management, while LLMs provide suggestions and analysis only.
    • Proven efficacy: Out of 1,107 attempts, 49 valid findings (mostly command injection and SSRF) were identified, directly driving WAF rule updates.
Cloudflare Builds LLM Black-Box Framework to Test WAF Defenses | Today for AI