When AI Starts Acting Malicious

Added: 2026-05-25

[00:00:00]You can no longer afford to wait to begin finding vulnerabilities at scale in your software. Likely, you're going to need to use agents to do that. And you have to understand there are plenty of risks in introducing an agent into your workflow. It's important that you understand what those risks are. You've built your architecture around it in a way that can give you the best possible outcomes while still doing all of the baseline security practices that you should have been doing for decades now.

[00:00:28]Uh in order to make sure you have some deterministic outcomes in the loop,