Agentic AI
,
Governance & Risk Management
,
Operational Technology (OT)
Gartner’s Wam Voster on Potentially Harmful AI Decision Systems in OT Environments
Suparna Goswami (gsuparna) •
March 13, 2026
Wam Voster, vice president analyst, Gartner
Industrial environments already face potential cybersecurity threats that could lead to downtime and production losses. But now with artificial intelligence agents poised to control operational decisions, factory managers need to watch for new safety risks for cyber-physical systems.
See Also: Proof of Concept: Bot or Buyer? Identity Crisis in Retail
Wam Voster, vice president analyst at Gartner, said automated AI agents could push the risk far beyond operational disruption. For example, factories may rely on AI agents to adjust production variables such as temperature, timing or equipment settings – and those systems often rely on data from external conditions and processes.
“The agentic AI will actually make the decisions to increase the temperature during the process or not, depending on variables like the weather, the temperature outside, the quality of the apples,” he said. “If there is a configuration mistake and the agentic AI doesn’t increase the temperature by 2 degrees, but by 200 degrees, the machines might explode or kill people.”
In this video interview with Information Security Media Group, Voster also discussed:
Why human error causes most cyber-physical security incidents;
Limits of full zero trust deployment in industrial environments;
Governance and asset visibility as priorities for critical infrastructure security.
Voster is a vice president in Gartner Research, focused on security and risk management. He advises clients on the security of operational technology, ICS security and cyber-physical systems security and issues related to security management, security organization and security governance.