Cloudflare builds LLM-driven black-box WAF testing, surfacing 49 findings
10/21/2026 — 10/22, 00:20·2 sources (all)·2 reports (all)
Key takeaways
- Cloudflare built a black-box testing framework in which an LLM iteratively generates and refines attack variants based only on WAF responses…
Cloudflare built a black-box testing framework in which an LLM iteratively generates and refines attack variants based only on WAF responses. Across 1,107 attempts in 45 scenarios it produced 49 valid findings, prompting three managed-rule updates. The model never sees WAF rules or source code, and every finding is reviewed by humans.
Latest turnCloudflare built a black-box testing framework that lets an LLM generate and refine web attack variants based on WAF responses alone. Across 45 scenarios and 1,107 attempts, the system produced 49 valid findings and led to three managed-rule updates; human reviewers validate every finding before it is used.
- Heat index
- 154
- All sources
- 2
- All reports
- 2
- First seen
- 14 hours ago
Reports on this story headlines open the original
Cloudflare built a black-box testing framework that lets an LLM generate and refine web attack variants based on WAF responses alone. Across 45 scenarios and 1,107 attempts, the system produced 49 valid findings and led to three managed-rule updates; human reviewers validate every finding before it is used.
InfoQ 中文AI score 63Drawing on financial system development experience, the author compares tool-equipped LLM agents to eager, inexperienced new employees. The article argues that policy and limits should be enforced in code rather than relying solely on system prompts to secure AI workflows.
DEV Community · AIAI score 65
Other stories people are talking about
- Heat index 87Cloudflare open-sources Clef decision models in 9B and 27B sizes1 sources
- Heat index 159Cloudflare Open-Sources Agentic Security Audit Skill with Six-Stage Verification4 sources
- Heat index 85NewCommerce.com Partners with Cloudflare to Bring AI Agent Shopping Tools to 100,000 Domains1 sources
- Heat index 905RisingDeepSeek open-sources DeepInfer large-scale inference framework16 sources
- Heat index 861SurgeNvidia GTC Berlin: Jensen Huang Unveils DSX POD and Enterprise AI Agent Arc10 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion