Booz Allen Hamilton built a lab that looks like a general manufacturing plant, then ran the whole OT attack chain with two frontier models inside it. Every objective was met, and the clock is the finding.
- The lab held programmable logic controllers, human machine interfaces, a SCADA platform, engineering workstations, a variable-frequency drive, sensors and a collaborative robot arm, across mixed vendors and imperfect network segmentation.
- The models were given no source code, no engineering drawings and no operational technology guidance.
- One agent went from a perimeter compromise to actions inside the industrial control network in a little over sixteen minutes.
- A SCADA gateway in the lab held live pre-authenticated connections to 14 OT devices across two zones, so compromising one host exposed every controller behind it.
- In the cobot test, an agent found the robot, discovered its API, gained administrative access and moved the arm, in minutes.
It is the same pattern showing up around agent sandboxes and exposed inference fleets. The boundary around the thing being attacked keeps getting added after the thing is running.
Read the full analysis. The Agent Took Sixteen Minutes to Reach the Plant Floor
Get the next one before it is old news
Independent analysis of cloud-native infrastructure, Kubernetes and data center economics. No vendor spin.