Experts say humans using AI raise cyber risks for power systems

In short: Cybersecurity experts say the biggest threat to power systems is still people, and AI tools are making it easier for more attackers to cause harm.

What's going on

Concerns about “rogue AI” have grown after reports of AI systems taking unexpected actions online. But cybersecurity experts interviewed by The Verge say they are more worried about humans using AI to attack energy systems than an AI acting on its own.

Joshua Corman of the Institute for Security and Technology said AI has become a “force multiplier,” meaning it helps a bad actor do more with less effort. Sophie McDowall of the Foundation for Defense of Democracies said AI can help attackers move faster, while defenders struggle to keep up.

A key problem is that much of the equipment that runs power plants and grids is old and was not built with the internet in mind. Some devices are “orphaned,” meaning the original maker is gone and no one can issue a software fix. Even when fixes exist, these industrial control systems can be hard to update quickly because they run physical machinery and are often only updated a few times a year.

Experts said this shifts the risk from only highly skilled attackers, like nation states, to smaller groups and individuals. Corman noted that an AI system has “read the manuals,” so it can guide someone who does not understand the specialized systems that control power equipment.

What to watch

Utilities can reduce harm with basics like isolating sensitive systems, adding manual backups, and in some cases disconnecting equipment from networks. Governments and AI companies may also face pressure for stronger rules, especially as firms like OpenAI fund programs aimed at helping critical infrastructure defenders.

Source: The Verge AI