Stop attacks before they are known: making the Cloudflare WAF smarter

What this post added

Introduced a WAF attack scoring system using machine learning to classify requests by their probability of being malicious. This system complements existing signature-based rules by detecting evasion, bypass, and new attack techniques before they are publicly known. The model is trained on millions of true positive samples and is optimized for SQL Injection, XSS, and RCE attacks. Also launched new Security Analytics to visualize attack score distributions and facilitate the creation of WAF Custom Rules.