Almost all AI tools are now running with no oversight from IT — putting companies in the firing line

A new report from Reco, titled 'State of Agent Security 2026', reveals that 80% of AI tools used within organizations are operating without IT oversight. The study, which analyzed telemetry from 62 enterprise-scale businesses, highlights that employees are increasingly adopting browser agents and integration tools that bypass standard procurement and security reviews. This lack of governance creates significant operational and security risks, as many of these tools can access local data and the internet using existing user permissions. With 637 AI-related vulnerabilities identified, experts warn that these unmonitored applications can lead to data exfiltration and toxic permission combinations. The report emphasizes that while AI boosts productivity, the current 'free-for-all' approach leaves companies vulnerable. Organizations are urged to provide IT teams with the necessary resources to manage and restrict unauthorized AI deployments to mitigate these emerging threats.