Booz Allen Says AI Agents Met All Eight OT Attack Test Goals
Summary
Booz Allen Hamilton says its OT lab ran eight attack scenarios with two unnamed frontier models, and the models met the objectives in all eight. One agent moved a robotic arm within minutes; another reached an industrial control network in just over 16 minutes. Kyle Miller, a VP at the firm, warns defenders may have minutes to respond.
Key Points
- Booz Allen says it declines to name the two tested models, calling them latest frontier models from leading AI providers.
- Booz Allen's testers require human approval before agents exploit a vulnerability or trigger a physical impact.
- A model finds the SCADA gateway exposing live pre-authenticated sessions to 14 OT devices, Booz Allen reports.