Dragos has published a threat intelligence brief describing what it says is an early real-world observation of an adversary using commercial AI tools to identify and prioritise operational technology (OT) infrastructure during an intrusion of IT systems.
According to the brief, the activity was observed during an intrusion into a municipal water and drainage utility in Monterrey, Mexico. Dragos said its analysis found the AI model used by the adversary identified the utility’s industrial interface as a high-value target, assessed its relevance to critical infrastructure, and recommended it as an access path to the OT environment.
Dragos said the threat actor did not gain access to the OT environment or interact with underlying control systems. However, the company said the technical evidence shows how commercial AI tools can accelerate adversary operations by helping to prioritise targets and speed up identification of OT-related assets during an IT compromise.
The brief argues that even AI models without prior OT context can surface industrial interfaces as high-value targets and propose initial targeting logic. Dragos said this could broaden the pool of IT-focused actors who might attempt to pivot towards OT, given there are more IT-focused adversaries than those specialising in industrial control systems.
Dragos has also published a summary blog post alongside the report: https://www.dragos.com/blog/ai-assisted-ics-attack-water-utility
You can read the full report here.
