Shadow AI: Power Users Pose Major Enterprise Risk

Enterprise AI risk is concentrated in a small group of power users, not spread broadly across staff. These top 5% of users run AI at 12 times the rate of other employees. Here's the thing: nearly half of all enterprise AI conversations happen through personal accounts, making them invisible to security teams. What's more, these power users often give work to AI agents operating outside company guardrails. A significant concern is that 16.31% of AI browser extensions have known vulnerabilities, and most of these request high-level permissions. Also, 14.4% of conversations use corporate email with personal freemium subscriptions, potentially exposing sensitive data to public model training. The bottom line: a small number of highly active AI users are creating the biggest security and privacy risks for businesses.