STAMFORD, Feb 12: Gartner, Inc., a leading business and technology insights company, today warned that misconfigured AI in cyber-physical systems (CPS) could trigger shutdowns of national critical infrastructure in a G20 country by 2028.

CPS includes operational technology (OT), industrial control systems (ICS), industrial automation systems, IIoT devices, robots, drones, and other AI-powered systems interacting with the physical world. Gartner highlights that even well-intentioned engineers, flawed updates, or minor configuration errors can result in large-scale service disruptions, posing risks to public safety and economic stability.

“The next major infrastructure failure may not come from hackers or natural disasters, but from AI misconfigurations,” said Wam Voster, VP Analyst at Gartner. “Implementing secure ‘kill-switches’ or override modes, accessible only to authorized operators, is essential to safeguard critical systems.”

Gartner notes that misconfigured AI can autonomously misinterpret sensor data or trigger unsafe actions, potentially impacting key services such as power grids, manufacturing plants, and transport networks. With AI models increasingly opaque, human intervention remains essential.

Gartner recommends key measures for mitigating AI-related risks in critical infrastructure:

  • Safe Override Modes: Secure human-accessible controls to retain ultimate authority over autonomous systems.

  • Digital Twins: Test configuration changes in virtual replicas before deployment.

  • Real-Time Monitoring: Continuous observation and rollback mechanisms for AI updates, supported by national AI incident response teams.

Gartner clients can explore these insights in Predicts 2026: Emergent Critical Risks of AI in CPS Security, and learn infrastructure modernization strategies via the complimentary Modern Infrastructure: Built for Tomorrow’s Business Demands roadmap.

Gartner continues to provide independent AI research and guidance for C-level executives, helping organizations safely leverage AI to achieve mission-critical priorities.