AI is outpacing security in critical infrastructure, executives and researchers warn

At CrowdStrike's Fal.Con conference, David Cooper, Americas cybersecurity growth leader at Ernst & Young, told theCUBE, SiliconANGLE Media's livestreaming studio, that the future of cybersecurity lies in humans and AI agents operating together at a speed not seen before. "There's a recognition that the future of cybersecurity is humans and agents operating at a speed we've never seen," Cooper said. He said his clients are all looking at how to help defenders accelerate and gain parity with attackers.

Payal Thakkar, EY's Americas industrials and energy cybersecurity leader, said oil fields, rigs and utility plants that operated unconnected for decades are now being tied into operational technology networks, widening the attack surface as agentic AI is introduced. "They are getting more and more connected into devices that were just not meant to be connected before," Thakkar said. She said the security model for those devices has to change, not just have an AI layer added.

EY research cited by SiliconANGLE projects that the share of organizations spending at least a quarter of their cybersecurity budget on AI will climb from 9 percent today to 48 percent in two years. Cooper said agentic AI is not replacing security operators, but the nature of their jobs is changing, and many clients still struggle to find staff with the right skills.

Separate research from Forescout, reported by TechRadar, found that AI-assisted attacks on operational technology are possible but not cheap or easy. Researchers used AI to port a remote-code-execution vulnerability from one programmable logic controller to another, triggering a denial-of-service state and executing attacker-supplied ARM shellcode. But the process required significant researcher input, consumed more than $500 in API usage, and a further attempt to extend the exploit bricked the PLC.

Forescout concluded that the difficulty, cost and specialist expertise required are likely to make such attacks less attractive than easier alternatives, TechRadar reported. TechRadar noted that nation-state attackers were not included in that assessment, pointing to Russia's Sandworm group hitting Poland's electricity suppliers in 2025 as a sign that resource-rich adversaries already target industrial devices.

CNBC reported that OpenAI Chief Executive Sam Altman said AI adoption is "non-negotiable" for countries. During a fireside chat with U.S. Commerce Secretary Howard Lutnick at the G20 Innovation Ministerial in Chapel Hill, North Carolina, Altman compared rejecting AI to refusing electricity more than 100 years ago. "The economic growth and benefit to people that can come from this, the value to a country, is too high to ignore," he said.

Altman predicted that children growing up today will never have known a world without smart, capable products and services. He described cybersecurity as one of the biggest challenges in the AI age. "I think some things are going to go very wrong with cybersecurity unless people act quite urgently," he said, warning that falling short in cyber defense could "set this technology back a great deal."