Day-Zero: An Adversarial AI Experiment

This is our third adversarial experiment.
Our first event was machines paying machines. A handful of developers built honeypots and tried to hijack the experiment. That's what told us the real story wasn't agent commerce, it was what happens when talented people point agents at each other.
That experiment became The Penthouse Heist.
Penthouse Heist put 125 people in a room with live infrastructure. 90 submissions. 39 canonical vulnerability families. Five distinct AI attack architectures converged on the same weaknesses within two hours. We learned that agentic systems don't fail in a hundred different ways, they fail in the same ways, at machine speed, over and over.
The Infrastructure Strikes Back gave defenders a seat at the table. Red vs. Blue. Hot-fix windows. A six-category defensive rubric. The conversation shifted from can you break it to can you break it faster than someone can fix it. And what happens when your telemetry might be lying to you?!?
Round 3 goes further.
New format. New targets. Same thesis: the only way to understand where agentic systems fail is to let talented people attack them in the open, at machine speed, with real infrastructure on the line.
Full brief drops soon. If you've been to a Day Zero round, you know what to do.