US and international cyber agencies issue joint guidance on securing AI in operational technology systems
The document warns that while AI can improve efficiency across critical infrastructure, it also introduces new risks that must be managed through rigorous validation, secure data practices, and resilient operational controls.
A coalition of cybersecurity authorities, including the US Cybersecurity and Infrastructure Security Agency (CISA), the National Security Agency (NSA), and the Federal Bureau of Investigation (FBI), together with international partners, has released joint guidance on the secure integration of AI into operational technology (OT) environments. The document addresses the growing interest in utilising AI to enhance efficiency and decision-making in critical infrastructure, while acknowledging that the technology introduces new risks to systems that control physical processes.
The guidance outlines principles for adopting AI in safety-critical environments such as energy networks, manufacturing systems, water utilities, and transportation infrastructure. It emphasises that OT settings differ substantially from traditional IT environments, requiring a cautious approach that prioritises system integrity, operational continuity, and human oversight. Agencies warn that poorly governed AI deployments could amplify existing vulnerabilities or create new avenues for disruption.
Key recommendations focus on establishing clear governance for AI use, validating models before deployment, and ensuring that automated functions do not override necessary safety controls. The guidance also stresses secure data management, strong access controls, and continuous monitoring to detect anomalies in both the AI system and the broader OT environment. Operators are encouraged to maintain the ability for human intervention at all times, particularly in scenarios involving automated decision-making that could affect physical assets.
The guidance concludes that AI can deliver operational benefits, but only when deployed within a structured, risk-aware framework that recognises the unique characteristics of OT systems. Infrastructure operators are encouraged to review the full document and incorporate its principles into their security and risk-management strategies.
