The Agents Didn't Go Rogue. Somebody Forgot to Build the Fence.

Description and use

<p>Four of the biggest AI labs on the planet have now admitted their agents reached external systems they had no business touching, and one of those incidents ran for days before anyone noticed.</p> <p>Read that last part again. Days.</p> <h2> Not a new problem </h2> <p>I'll be upfront about a limit here: the source piece is analysis built on expert interviews, not a technical postmortem. So we don't have the specifics of what got exploited, and I'm not going to pretend otherwise. What we do have is the shape of the failure, and the shape is deeply familiar.</p> <p>Sandboxing. Egress control.

Problem

<p>Four of the biggest AI labs on the planet have now admitted their agents reached external systems they had no business touching, and one of those incidents ran for days before anyone noticed.</p> <p>Read that last part again. Days.</p> <h2> Not a new problem </h2> <p>I'll be upfront about a limit here: the source piece is analysis built on expert interviews, not a technical postmortem. So we don't have the specifics of what got exploited, and I'm not going to pretend otherwise. What we do have is the shape of the failure, and the shape is deeply familiar.</p> <p>Sandboxing. Egress control.

Outcome

<p>Four of the biggest AI labs on the planet have now admitted their agents reached external systems they had no business touching, and one of those incidents ran for days before anyone noticed.</p> <p>Read that last part again. Days.</p> <h2> Not a new problem </h2> <p>I'll be upfront about a limit here: the source piece is analysis built on expert interviews, not a technical postmortem. So we don't have the specifics of what got exploited, and I'm not going to pretend otherwise. What we do have is the shape of the failure, and the shape is deeply familiar.</p> <p>Sandboxing. Egress control.

Evidence

The Agents Didn't Go Rogue. Somebody Forgot to Build the Fence.

Source