Why Trustpilot treats AI agents as the ultimate insider threat - and what that looks like in practice
Let's remove anthropomorphism from it. They don't think.
That was one of the first things Dan Conn said when I sat down with him at Chainguard Assemble in London in October. The Trustpilot Staff Platform Security Engineer had just come off a panel where he'd called AI agents "the ultimate insider threat." He elaborated:
It's something I should probably clarify. Agent systems are a little bit more sophisticated than usual automation processes. They don't think - but they have behaviors that allow them to work in a way that is a bit unusual compared to our normal deterministic systems. They're going to do things that maybe an intern might do if they were in our system. So we have to think around them as an insider threat, as in the same way that we should think really about employees as an insider threat.
He pointed to the Hugging Face incident...
Copyright of this story solely belongs to diginomica.com. To see the full text click HERE
