The Use and Governance of AI Agents at U.S. Firms

Autonomous AI agents have been forecast to change work and organizations, but evidence about firms’ deployment of these tools is scarce. Drawing on a survey of 1,500 full-time senior managers, we provide some of the first systematic evidence about the diffusion of AI agents, defined as tools able to take actions and complete tasks autonomously. Just over half of all managers surveyed report that their organization has at least one AI agent built into its work processes, but fewer than half of managers who reported this are aware of a formal policy that governs the use of agents at their organization. The correlation between integrating AI agents into work and enacting policies to govern them is positive but small, suggesting that integrating and governing agents might be separate sets of choices made by organizations. These results suggest that while AI agents are a real phenomenon in US workplaces, organizations vary in how they are adjusting work and policies in response, which may affect agents’ ultimate impacts.