In this TechTalk, I kick things off the only way I know how: by joking that we “don’t use AI” while I’m literally in the middle of setting up an AI server. Then we go full doom-mode on what AI is doing to security right now, because if you’re not updating your stuff constantly, you’re basically volunteering to be a case study. The big takeaway: AI doesn’t just help find bugs faster—it can turn around and generate working exploit code just as fast, and that completely changes the timeline from “bug found” to “you’re owned.” I break down a nasty real-world example: hundreds of GitHub Python projects getting infected with malware that hid inside strings using invisible Unicode characters that don’t render in most editors. It looks like an empty string, but it’s actually a giant encoded payload that later gets decoded in-memory and executed without ever writing a file—so traditional file-based scanning (and even tools like VirusTotal in some cases) can miss it. And the truly terrifying part is AI can mutate the encoding and decoder per project, so signatures don’t help much. My advice is simple: stop running random GitHub code blind, and start treating updates, scanning, and code review like it’s daily hygiene—because it is now.