Global cybersecurity agencies released new guidance on safely deploying AI in operational technology systems, citing the high risks involved. OT is essential for critical infrastructure, and integrating AI can introduce risks such as model drift, safety bypasses, and process instability. Agencies urge thorough education on AI risks, a careful assessment of when to use AI, strong data controls, and transparent governance. Recommendations also emphasize the importance of monitoring and fail-safe processes, including human oversight. Experts note that while AI may enhance efficiency, its use in OT should be limited and highly disciplined, especially with high-risk models like large language models.
https://www.darkreading.com/cybersecurity-operations/cisa-publishes-security-guidance-ai-ot