45% of execs limit human AI oversight to high-stakes work—or don't have any oversight at all

A Zapier-commissioned survey of 518 U.S. executives at companies with formal AI governance policies finds wide variation in how organizations oversee AI-powered work, despite significant concern about autonomous systems making flawed decisions. While 85% of respondents report some human approval process, 30% reserve intervention for high-stakes actions, 29% approve most actions, 26% review every feasible action, and 15% allow fully unsupervised autonomy. More than one-third of executives said unreviewed AI had already caused negative consequences, including rework, lost or delayed revenue, reputational harm, employee discipline, and legal or compliance issues, yet 49% personally trust AI to make most decisions without review. The findings suggest that effective human-in-the-loop governance should prioritize regulatory requirements, whether actions can be reversed, financial or business impact, and departmental risk rather than requiring universal approval, which may create inefficient rubber-stamping. The survey, fielded by Centiment for Zapier in July 2026, argues that organizations need clear, targeted oversight and tools that document approvals and control AI access to company systems.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.