In this TechTalk, I kick things off the way we always do: a little chaos, a little “we do what we want,” and then we dive into the stuff that actually matters. One of the big conversations in this episode is Discord rolling out age verification and “teen by default” settings. On paper they’re saying it’s not for everybody, but the reality is it’s still a privacy landmine. They’re talking about facial age estimation, uploading IDs through vendor partners, and even “age prediction” using info they already have—which honestly is even creepier, because it implies they can profile you without you ever explicitly consenting to hand over anything. I also dig into why this kind of system is easy to game (virtual cameras, AI-generated faces, using a parent’s ID) while still punishing legitimate users who look young or get misclassified. And then we zoom out into the bigger picture: AI-driven security risks. If AI can help developers find bugs, it can absolutely help attackers find and exploit them too—prompt injection, model poisoning, reverse engineering, agentic AI grinding on an exploit until it gets purchase. The takeaway is simple: we’re way more exposed than people want to admit, and the next wave of cyber threats is going to be AI-assisted whether companies are ready or not.