Security, governance, and accountability for autonomous agents. · 55 articles
Most agent breaches don't happen because the model "went rogue." They happen because a well-behaved agent was handed a credential that could
A trust-and-safety (T&S) function inside an Agentic AI-as-a-Service company is not a support desk with a fancier name. It owns the rules an
A confused deputy is a program that has legitimate authority and gets tricked into misusing it on someone else's behalf. Tool-using AI agent
Sandboxing an AI agent means giving it a controlled box to operate in so that when it misbehaves -- and at scale, it will -- the blast radiu
When an AI agent does something nobody told it to do, the clock starts immediately and the wrong instinct is to debate philosophy. This play
Agentic AI introduces a leak surface that traditional breach playbooks don't cover: an autonomous agent can exfiltrate sensitive data throug
When you buy an AI agent as a service, you're not just renting a worker. You're handing a piece of software a set of keys to your systems, t