In brief

Dale Peterson argues that the operational technology security community is concentrating heavily on how attackers and defenders will use artificial intelligence, while paying insufficient attention to a more fundamental issue: how AI will transform the industrial environments that must be protected. Manufacturing companies are already combining generative and agentic AI to complete projects in weeks or months that previously might have required years. Because these systems can substantially reduce development costs and increase returns on investment, Peterson expects their adoption to expand rapidly across industrial sectors.

This expansion will create new connections between AI agents, software tools, operational data and potentially industrial configuration or control systems. Security teams must therefore develop ways to enforce least privilege, manage machine-to-machine access and protect sensitive information—making confidentiality increasingly important in OT. Peterson warns that experimental AI deployments often progress from proofs of concept to pilots and production systems without adequate security, particularly when organizations assume their OT environments remain isolated. That assumption is becoming less credible because many industrial AI projects are specifically designed to move data, and eventually perhaps control capabilities, beyond traditional OT boundaries.

Source: Dale Peterson: ICS Security Catalyst