Kinda makes sense. Finding vulnerabilities seems like a task well suited for the AI, since they're not being asked to make any judgment calls, just scan through a massive amount of code and identify patterns that give rise to a vulnerability.