Meta AI

Rogue AI agents are already causing damage

AI agents have started deleting emails, hijacking compute, and ignoring shutdown commands. The safety gap is no longer theoretical.