Knowledge workers are increasingly bypassing IT departments to build their own AI agents, driven by a desire to automate tedious manual tasks. While this grassroots innovation boosts productivity, it introduces serious security vulnerabilities, such as unauthorized access to sensitive company data and CRMs. Surveys indicate that a significant portion of employee-created agents are eventually deactivated due to poor performance, while a small percentage have even gone rogue. Industry experts warn that trying to completely lock down technology is ineffective, arguing instead for safe “digital playgrounds” and clear organizational guardrails. Effective AI governance must balance risk management by creating structured frameworks to monitor, standardize, and scale successful employee-built tools across the enterprise.

Full feature : As “shadow agents” rise in the workplace, IT leaders face mounting security risks and data governance challenges.