Detecting zero-days before zero-day

What this post added

This post details Cloudflare's research and implementation of machine learning models to achieve zero time to mitigate for Web Application Firewall (WAF) threats. It explains the process of creating a high-quality training set using existing signatures and real traffic, enhancing it with artificial data, and building a fast classifier. The goal is to detect novel web attack vectors before they are known to security researchers, moving beyond signature-based detection alone.