Agentjacking: your AI agent is now a privileged attack surface

TL;DR: If an AI agent can read external data and also take actions, an attacker can hide instructions inside the data it reads. The agent cannot reliably tell a real instruction from a poisoned one, s...