Inside AI-Powered WAF Detections: Architecture and Safety Controls
Akamai, Tuesday, September 8th, 2026
How AI-based web application firewall detections are built and what stops them blocking legitimate traffic.
This post details the architecture behind Akamai's AI-powered WAF detections and the safety controls that prevent them from blocking legitimate traffic.
The safety design is the substantive content: a WAF that occasionally blocks valid requests is worse than one that occasionally misses attacks, because the business impact is immediate and visible, which is why teams run WAFs in detection-only mode for years. Understanding how a vendor constrains model-driven blocking decisions is the right basis for deciding whether to enable enforcement.