CISA Publishes Security Guidance for Using AI in OT

**Executive Summary:** The joint guidance from CISA, FBI, NSA AI Security Center and allied international agencies warns that integrating AI (especially LLMs/GenAI) into operational technology (OT) creates significant safety and security risks—including model drift, hallucinations, data leakage, and potential RCE—and recommends that OT operators prioritize foundational cybersecurity, establish AI governance and assurance frameworks, minimize model data access, perform rigorous testing and evaluation, and embed human oversight and fail-safe mechanisms to protect critical infrastructure and ensure safe, reliable operations.