What AI Actually Changes for OT Security: Observations from the Field
Summary: Dragos reports that after nine months of first-hand work with frontier AI models, it has seen adversaries use AI in early-stage operations, including a breach of the Mexican government where the model unprompted identified a compromised system as an interface to a government-owned water utility and supplied vendor names, default configurations, and default passwords. Dragos also says the models found dozens of strong zero-days across major OEMs plus hundreds to thousands of additional vulnerabilities in industrial products, and that a threat group it tracks cut exploit development from 24 to 48 hours toward minutes with AI assistance.
What you should know: Dragos estimates fewer than 10 percent of OT environments worldwide have visibility and monitoring in place, while some 90 percent of owners and operators run prevention-only postures and more than 80 percent of standards guidance is preventative - so most environments have no way to tell whether an event was a cyberattack when prevention fails.
Read the original at Dragos Blog ↗
- Topics
- OT / ICS & Critical Infrastructure
- Related
- Siemens RUGGEDCOM APE1808 Cyber Risks · Nov 12, 2026How to Assess OT Incident Response Preparedness Cyber Risks · Nov 10, 2026Three China-Nexus Espionage Campaigns Linked Through Sparrow Malware Ecosystem Cyber Risks · Nov 10, 2026