More than half of employees use unapproved AI tools at work

KnowBe4 reveals 55% of UK workers admit to using unapproved AI tools
Only 16% believe they are effectively managing safe AI.
27% complement enterprise AI with more tailored tools

More than half (55%) of UK employees admit to using unapproved AI tools at work, and up to one in ten knowingly share sensitive company information with these unauthorized tools, according to new research.

KnowBe4’s new report aims to define shadow AI as untrusted AI, not use of AI that goes unnoticed, as the number of UK cybersecurity decision-makers identifying shadow AI as the biggest risk is almost equal (58%), implying they are well aware of the challenges.

However, little appears to be being done, as only 16% believe their organization is currently effective in managing the safe use of AI.

Shadow AI is an ongoing challenge

Nearly half (46%) have implemented goals to improve AI agent safety over the next 12 months, but with one in five (19%) already reporting that AI agents are taking autonomous action across multiple workflows with limited human oversight, the risks remain clear.

“UK businesses are adopting AI to improve productivity [but] Many employees are still under pressure, using unapproved tools, and regularly encounter (and fear) sophisticated threats like deepfakes and phishing,” wrote Javvad Malik, Chief CISO.

Shadow AI doesn’t necessarily mean employees are turning their backs on enterprise-grade tools, as 27% admit they occasionally source their own tools in addition to those provided to them, indicating that companies are failing to provide the right tools workers need.

So while governing AI use with clearly defined policies is an area for improvement, simply providing workers with the tools they need could go a long way toward reducing the impact of shadow AI on any organization.

Google logo on black background next to text