Thought Leadership
According to a Honeywell study, most OT security teams are already using AI, but autonomous systems remain the exception.
Honeywell has published its 2026 OT Cybersecurity Benchmark Report. The results show a discrepancy between industrial companies’ self-assessment of their operational technology (OT) security and their actual readiness. The report also examines what impact AI is already having on OT security.
Gap Between Self-Assessment and Readiness
Among the 603 surveyed security leaders from critical infrastructure sectors, 88 percent described their OT security program as mature or at least structurally well-conceived. At the same time, only 21 percent have a complete inventory of their OT devices. Leaders from sectors such as energy, oil and gas, healthcare, maritime, and manufacturing across the Americas, EMEA, and Asia-Pacific regions participated in the survey.
Significant gaps were also evident in monitoring: Only 33 percent of respondents stated that OT systems are fully integrated into a central Security Operations Center, and merely 20 percent continuously monitor more than three-quarters of their connected IoT devices. In healthcare, only 19 percent reported that building and facility systems are fully integrated into security monitoring.
Substantial Costs from Downtime
Organizations that experienced a major OT security incident reported an average of 16.2 hours of downtime. Among affected respondents, 21 percent estimated downtime costs at more than 100,000 US dollars per hour, while 4 percent even put losses at more than 500,000 US dollars per hour.
The frequency of incidents varied significantly depending on the industry. 91 percent of respondents from the energy and utilities sector as well as 87 percent from maritime reported a major OT security incident within the past twelve months, compared to 54 percent in the oil and gas industry.
AI Already Widespread, Autonomy Rare
99 percent of respondents expect AI to impact their OT security operations within the next two to three years. AI-powered tools are already widely used in OT security: 72 percent of respondents use AI for threat detection, 68 percent for continuous monitoring, and 59 percent for device inventorying. By contrast, only 23 percent currently deploy autonomous or agent-based AI for threat detection, indicating that most systems currently assist human analysts rather than act independently. Honeywell writes in the report:
“As AI moves from merely assisting analysts to taking autonomous action, organizations require clear decision-making authority, human oversight, and testing that accounts for the operational consequences of a flawed response. The goal of well-governed AI automation is to strengthen visibility and responsiveness without creating new risks to uptime, assets, or safety.”
Honeywell
(Editorial Team)
