AI-Powered Cyberattacks Are Becoming a Bigger Threat to Energy Infrastructure
AI is lowering the barriers to sophisticated attacks on energy and other critical infrastructure, turning what were once specialised operations into more accessible campaigns. Recent alerts from US and European agencies show that threat actors are already using AI-generated scripts and tools against industrial control systems.
In August 2026, the NSA, CISA, FBI, Department of Energy and EPA warned that hackers were targeting Siemens S7 series programmable logic controllers with AI-generated exploitation scripts. The same controllers appear in water, energy, chemical and manufacturing facilities. Officials noted that AI dramatically reduces the expertise and time required to develop working industrial-control exploits and allows attackers to adapt more quickly to defences.
Similar patterns have appeared elsewhere. Researchers documented an AI-assisted intrusion attempt against a Mexican water utility in which commercial models helped map networks and identify pathways into operational technology environments. Iran-linked actors were reported to have disrupted a UK power facility, while other campaigns have targeted energy networks in Europe and Latin America. Although many of these efforts have so far failed to cause widespread outages, they demonstrate active capability development.
The risk is structural. Modern power systems rely on software for generation control, frequency regulation and automated switching. AI-enabled attackers do not need to cause a total blackout to inflict damage; subtle manipulation of control logic or power quality can disrupt sensitive loads such as data centres or industrial processes. At the same time, the rapid expansion of renewable generation and AI-driven electricity demand is increasing the complexity—and the attack surface—of the grid.
Defenders are responding with AI of their own for anomaly detection, threat hunting and simulated attack testing. Yet the asymmetry remains: offensive tools can be generated and iterated rapidly, while securing legacy industrial systems is slow and expensive. Energy operators, regulators and security firms now treat AI-assisted attacks as an active rather than theoretical threat to critical infrastructure.
