Making industrial AI safer as it moves into physical automation

Industrial AI is transitioning from basic predictive analytics toward more sophisticated automation driven by agentic and physical AI models. According to Arti Garg, chief technologist at AVEVA, these advancements allow for the management of intricate tasks within physical environments. Unlike purely digital software, these systems interact directly with hardware and critical infrastructure, meaning any unforeseen algorithmic decisions could impact operational safety and reliability.

This shift necessitates a focus on responsible deployment and data integration. Because newer AI models are often more difficult to interpret, businesses must find ways to correlate diverse information sources, such as engineering documents and telemetry, to ensure real-time situational awareness. As these technologies evolve, companies should prepare for potential changes in regulatory or compliance obligations. It is essential for organizations to verify exactly which new rules apply to their operations, identify effective implementation timelines, and establish rigorous documentation standards for proof of compliance.