Your AI Agent Can Be Socially Engineered. Here Are 3 Attacks That Prove It.

No jailbreak. No exploit. No alert fired. Just a conversation. In September 2025, a Chinese state-sponsored threat group ran a cyberattack against 30 organizations globally. The human operator clicked a button. The AI did the rest — reconnaissance, lateral movement, exploitation — autonomously, at machine speed, without writing a single line of code. This wasn't a proof-of-concept. This was an operational attack. And here's the part security teams don't want to hear: the same AI systems being de