Improving the WAF with Machine Learning

What this post added

Introduced a machine learning-based detection system to complement existing WAF managed rulesets. This system scores every request for the likelihood of SQLi or XSS attacks, providing a new WAF analytics experience. The model is trained on good/bad traffic classified by managed rules and augmented data, improving detection rates and identifying bypasses proactively. Mitigation is performed via custom WAF rules using the ML-generated scores. The system exposes scores for SQLi and XSS, with plans to extend to all managed rule attack vectors.