Why traditional access control fails for AI agents, and how relationship-based authorization provides a path forward.




The first time I watched an AI agent autonomously chain together twelve API calls to complete a task, I felt two things simultaneously: excitement at the capability, and dread at the security implications.

The agent had been...