IT managers report widespread lack of control over proliferating AI agents

12 articles · Updated · ZDNet · Apr 28

A Rubrik ZeroLabs survey finds only 23% of IT managers have full control, while 81% say agents require more manual oversight than expected and 86% foresee security guardrails being inadequate within a year.
Agent sprawl, driven by easy creation and bypassed security controls, is causing fragmentation, inconsistent governance, and hidden security gaps across enterprises, with most lacking rollback capabilities for unintended agent actions.
Industry experts warn that unchecked agent proliferation mirrors early cloud adoption risks, urging organizations to treat agent management as a first-class discipline to balance speed with effective governance and security.

Is the AI revolution creating more manual work and security risk than actual business value?

After the McKinsey AI hack, are AI prompts now the new corporate 'crown jewels' to protect?

How can firms apply Zero Trust security to thousands of AI agents that lack a clear identity?

Can new governance toolkits from Google and Microsoft truly tame AI, or just create vendor lock-in?

Are companies sleepwalking into massive fines with new AI laws arriving as soon as June?

Does 'harness engineering' for AI solve safety or just stifle the innovation it aims to protect?