The 5 petabytes of daily operational technology telemetry Dragos Inc. has been stockpiling for over a decade now has a new job: teaching an AI assistant how to tell the difference between a broken pump and a Russian hacker probing a power grid. That assistant, EmberAI, launched Monday. The immediate consequence is a shift in who can defend critical infrastructure.
For years, the shortage of OT security specialists has been a known, intractable problem. Attacks on power grids, water systems, and pipelines are climbing.
The pool of people who understand both industrial control systems and cybersecurity remains shallow. EmberAI does not create new specialists. It attempts to make the ones that exist — and the IT generalists and plant engineers who do not — far more effective.
The tool runs on the Dragos Intelligence Fabric. That fabric is the company’s proprietary data set, built from adversary tracking and incident response work spanning more than 600 OT protocols.
It also includes research from Dragos’s status as a Common Vulnerabilities and Exposures Numbering Authority. That volume of data is the foundation. Without it, the AI would be useless in an OT environment.
General-purpose AI, Dragos argues, cannot distinguish critical exposures from routine noise because it lacks understanding of the plant and its systems. EmberAI is designed to have that understanding baked in. The practical effect for an organization securing a manufacturing plant or a data center is straightforward.
An analyst can query assets, vulnerabilities, and network activity in plain language. The system ranks threats by operational impact.
A veteran OT specialist can use it to speed up analysis. An IT practitioner who has never touched a programmable logic controller can use it to ask what matters right now. A plant engineer can use it without a cybersecurity background.
The tool collapses the skill gap by embedding expertise into the interface. What to watch next is adoption speed among the sectors Dragos targets.
The company lists power grids, manufacturing plants, water systems, pipelines, and data centers as the critical infrastructure EmberAI is built to protect. These are not sectors that move fast. They are sectors where a mistake in the digital layer can cause physical destruction.
Trust in an AI assistant that recommends shutting down a line or patching a controller will not come overnight. The data set is enormous — more than 5 petabytes of daily OT telemetry — but trust is not built on data alone. It is built on outcomes.
Another consequence is competitive pressure. Dragos has positioned EmberAI as a tool specifically for OT, arguing general-purpose AI cannot handle the job.
Rivals in the industrial cybersecurity space will have to respond. Either they match the specialization or they concede that general models are insufficient for critical infrastructure. The skills shortage is not going away.
The attacks are not going away. The market for tools that can amplify the existing workforce is only growing.
EmberAI does not solve the shortage. It does not train new specialists. It does not fix the underlying problem of too few people who understand both the plant floor and the network.
What it does is let the people who are there work faster and smarter. For a water utility with one part-time security person, that might be enough. For a pipeline operator under active threat, it might be the difference between a contained incident and a blackout.
The tool is a force multiplier, not a replacement. That distinction matters.
{“@context”: “https://schema.org”, “@type”: “NewsArticle”, “headline”: “Dragos trains AI assistant on decade of OT data to counter skills shortage”, “description”: “The 5 petabytes of daily operational technology telemetry Dragos Inc. has been stockpiling for over a decade now has a new job: teaching an AI…”, “datePublished”: “2026-06-23T13:21:00+00:00T09:30:00+00:00”, “author”: {“@type”: “Person”, “name”: “Ava Anderson”}, “publisher”: {“@type”: “Organization”, “name”: “InfoPulse Today”, “url”: “https://newsfeed.infopulsetoday.com/”}}
