The new AI-driven AppSec tool reportedly uncovered hundreds of critical flaws and thousands of high-severity issues during early testing.