38% of risk leaders express concern over unintended actions from runaway processes, such as unauthorized transactions or incorrect pricing changes, as a risk from deploying agentic AI