AI agents

Ghostjacking: When AI Agents Read Poisoned Security Logs

A new attack called Ghostjacking hides instructions in security logs so AI agents execute them automatically. Here's how it works and how to defend.