Artificial intelligence is rapidly becoming both a defensive necessity and a strategic risk factor for critical infrastructure operators as cyberattacks increasingly spill into the physical world. Ricardo Villadiego, founder and chief executive officer at Lumu Technologies, wrote in a story published on the World Economic Forum, that nation-state actors and ransomware groups are targeting hospitals, utilities, power grids, schools, and transit systems that now rely heavily on connected OT (operational technology), internet-enabled sensors, and programmable logic controllers.
Villadiego argues that human-led monitoring alone can no longer keep pace with AI-assisted adversaries, particularly as telecommunications and education sectors account for nearly half of global ransomware targeting activity combined. With 76% of organizations reportedly taking more than 100 days to fully recover from cyber incidents, the operational consequences now extend far beyond data theft into real-world disruption of essential services.
“Defending critical infrastructure requires a comparable automated response and human teams cannot secure this environment alone,” Villadiego observed. “Adopting AI-driven network security remains the only practical way to manage this massive influx of automated risks.”
He argues that AI-driven attacks on critical infrastructure are becoming faster and more effective, making AI-powered defense essential to match the speed and scale of emerging threats. “Attackers use AI to move much faster than human teams can deploy patches. Because of this speed advantage, one in six successful data breaches now involves attacker-driven AI. Given this threat, 87% of organizations identify AI-related vulnerabilities as the fastest-growing cyber risk.”
Villadiego said that one may assume that AI means speed, but that is not always the case. “Adversaries often use this technology to prioritize stealth over immediate destruction. Recent threat intelligence shows defence evasion is the most dominant attack tactic today. Cybercriminals deploy stealthy campaigns that avoid triggering alarms by moving quietly over long periods – meaning that when they attack, they cause maximum impact.”
He noted that “Defending critical infrastructure requires a comparable automated response and human teams cannot secure this environment alone. Adopting AI-driven network security remains the only practical way to manage this massive influx of automated risks.”
Recognizing that modern critical infrastructure facilities rely on unmanaged physical devices, he noted that security teams cannot install traditional software agents on these internet-connected digital switches, medical equipment, water valves, or many other physical assets. This leaves an undefended attack surface for attackers to exploit.
Villadiego said that hardware limitations further complicate this defence strategy. Utility companies often cannot install advanced AI defences on even their traditional endpoints. Industrial equipment has a notoriously long lifespan and, consequently, much of this operational infrastructure still runs on outdated systems like Windows 95, NT, or Windows XP.
“The network provides the only viable answer. Cyber resilience across an OT landscape requires continuous monitoring at the network layer,” he pointed out. “The network sees every connection, so AI applied to network traffic catches the evasive behaviors that local endpoint tools completely miss. AI can instantly read millions of data points and analyse traffic patterns across both digital and physical landscapes, rapidly recognizing malicious anomalies. It then severs harmful connections before an advanced threat jumps into physical control systems.”
Critical infrastructure is becoming heavily digitalized, Villadiego said, adding that facilities now rely on a variety of sensors and smart meters and use internet-connected programmable logic controllers (PLCs), which act as a digital brain to control and automate physical machinery. “This massive scale of infrastructure networks makes manual monitoring impossible.”
He added that ransomware gangs are aware of this and heavily target this digital-physical infrastructure. “Recent data confirms this trend with telecommunications and education currently topping global targets, accounting for 24.8% and 23.4% of attacks, respectively.”
Citing IBM’s Cost of a Data Breach Report 2025, Villadiego pointed out that “76% of organizations take more than 100 days to fully recover from an incident. We are already seeing these devastating impacts in the real world.”
He also highlighted that recent attacks on the U.S. water and wastewater sector prove this vulnerability. “In late 2023, Iranian state-sponsored hackers successfully targeted Israeli-made PLCs at municipal water facilities in the US. They exploited internet-connected devices and locked facility operators out of the systems. The affected water authorities had to immediately take systems offline and switch to manual operations.”
Villadiego also highlighted the future of automated defense for critical infrastructure providers, as severe cyber talent shortages continue to leave state and local security teams heavily overburdened. “AI bridges this dangerous gap by acting as a vital force multiplier. The technology automates labour-intensive tasks like alert triage and log analysis. As AI brings accurate network intelligence to the surface at machine speed, human experts gain the freedom to confidently guide final operational decisions.”
“The attack surface continues to expand across digital and physical environments. AI-driven network defence provides the only scalable answer to this threat,” he explained. “Organizations can no longer rely on partial visibility provided from endpoints, so the network remains the only absolute truth in a compromised environment.”
Moreover, deploying AI in network defences is not just about stopping hackers; it is about building digital trust. “This trust is necessary to safely embrace the current industrial modernizations and ensure the uninterrupted delivery of vital services.”
Earlier this month, the U.S. Cybersecurity and Infrastructure Security Agency (CISA), alongside the Australian Cyber Security Centre and other international partners, published guidance on the secure adoption of agentic artificial intelligence (agentic AI), outlining cybersecurity risks tied to deploying these systems. The document comes as critical infrastructure and defense sectors increasingly adopt agentic AI to support mission-critical operations and drive automation. As agentic AI systems play a growing operational role, defenders must implement security controls to protect national security and critical infrastructure from agentic AI-specific risks.
