ITT warns that agentic AI systems may introduce operational, security, compliance, and other risks.
“Agentic AI systems are tools capable of autonomously initiating tasks, making decisions, or executing multi-step actions without direct human prompting, which may introduce additional operational, security, and compliance risks. These systems could take actions outside intended business parameters, propagate errors at scale, or interact with internal and external systems in unanticipated ways, increasing the likelihood of operational disruptions or regulatory exposure. Agentic AI may also heighten risks related to model alignment, permissions design, and dependency on third party control mechanisms, any of which may fail to function as expected. If agentic AI systems behave unpredictably, are misconfigured, or are exploited by malicious actors, they could result in unauthorized transactions, data leakage, safety incidents, or other adverse impacts. If we are unable to effectively govern or oversee these autonomous capabilities, we could experience material and negative affects to our business, reputation, financial condition, and results of operations.”