Company execs say employees are skirting AI rules
A Zapier survey of 548 U.S. executives finds 79% report employees bypass AI governance rules to deploy workflows faster, with 52% calling it widespread and enforcement uneven across company sizes.
A Zapier survey of 548 U.S. executives at companies with 500 or more employees found that 79% say workers are actively bypassing or working around AI governance processes to deploy or modify workflows more quickly. Only 21% report such behavior is rare or nonexistent. The issue appears most pronounced in mid-sized companies, where 81–82% of executives acknowledge rule-skipping, compared to 64% at firms with 5,000 or more employees.
Executives cite two primary approaches to AI deployment: 41% prioritize speed, launching workflows even if governance is refined later, while 47% wait until policies and testing are fully established. The lack of consensus within leadership contributes to employees bypassing rules, as policies documented in PDFs are ineffective when the executive suite itself disagrees on the strategy. This tension mirrors traditional software deployment debates but carries higher stakes, as ungoverned AI workflows can impact CRM systems, billing, and customer data simultaneously.
The consequences of ungoverned AI workflows are tangible, according to executives: 47% report increased manual workload or operational costs, 41% cite client-facing errors, and 26% note revenue impacts. Despite 85% of executives stating AI runs mission-critical workflows, enforcement remains inconsistent. While 80% say their organizations have formal audit processes, 28% admit policies exist only on paper and are not consistently enforced across teams or use cases.
Mid-sized organizations struggle most with enforcement, with 41% of executives at firms with 1,000 to 2,499 employees reporting inconsistent policy application. Only 4% describe governance as actively failing, but 49% acknowledge room for improvement. Meanwhile, 29% review governance quarterly or less often, highlighting a gap between policy and real-time oversight needs. Experts suggest governance must move at the same speed as AI tools, with real-time tracking and access controls to prevent disruptions in critical workflows.