Shadow AI Is Now an Enterprise Liability: Two-Thirds of Workers Use Unauthorized Tools Despite Warnings

A new study from PagerDuty has put a number on what many security and operations leaders have suspected: shadow AI is no longer a fringe behavior. Two in three office professionals admit to using AI tools or services their company explicitly prohibits — and the data exposure is significant.

The Numbers Behind the Risk

The PagerDuty report surveyed office professionals across industries and found:

• 66% have used unapproved AI tools at work, even knowing they were banned
• 53% received informal warnings to stop — and continued anyway
• 48% faced formal disciplinary action but persisted
• 88% have shared work-related information with public AI systems like ChatGPT, Claude, or Gemini
• 43% uploaded emails, 40% shared meeting notes, 34% entered customer information, and 31% shared sensitive business documents

The Underlying Tension

The root cause isn’t recklessness — it’s a gap between what workers want and what IT provides. Three-quarters of those surveyed (77%) said their company’s AI restrictions are limiting their professional development. A similar proportion of both individual workers and senior leaders believe they understand AI better than their tech teams.

PagerDuty CTO Tim Armandpour put the challenge plainly: the goal for executives shouldn’t be slowing AI adoption — it should be redirecting that energy into governed platforms that can scale safely.

What Organizations Can Do

Rather than blocking tools workers are determined to use anyway, experts suggest observing which AI tools employees gravitate toward and building enterprise-grade security on top of those — turning shadow usage into structured adoption before sensitive data walks out the door.

Read the full article on TechRadar