The news: IT leaders are increasingly concerned about unauthorized employee use of AI and its risk to company security and compliance.

• 90% are concerned about “shadow AI,” or employees adopting AI tools without IT team approval, per Komprise’s IT Survey: AI, Data & Enterprise Risk.
• 13% of companies said genAI has harmed their finances, customers, or reputation—proof that AI’s risks aren’t just hypothetical.

The damages: More than three-quarters (79%) said their company has been negatively affected by employees using generative AI (genAI).

• Almost half (46%) of damages were related to false results from AI use.
• 44% said sensitive company data had been leaked into AI models.

Zooming out: AI’s threat to jobs and creativity, and potential to create heavier workloads, are pushing employees to sabotage their companies’ AI initiatives. About one-quarter (27%) of employees have retaliated against their companies’ genAI strategy by entering company info into an unapproved AI tool, per Writer.

A lack of investment in employee training and resources can also lead to shadow AI use.

• 49% of executives said employees are left on their own to figure out genAI.
• Two-thirds (35%) of workers are paying out of pocket for genAI tools because their employer doesn’t provide the ones they want.

Taking action: Some of the ways IT leaders are addressing these risks include data management, employee monitoring, and establishing clear company policies regarding AI use.

• Three-quarters (74%) of IT leaders are tagging and securing sensitive data in their systems so it’s not available to be entered into AI systems, per Komprise.
• About the same number (73%) are tracking which AI tools are being used by employees, which employees are using them, and what company data sets have been fed into AI models.

Despite those efforts, only half (55%) are prioritizing employee training and policy development, meaning more workers are being monitored for their AI use than are being educated on what they may be doing wrong.

Our take: Companies should pair data management and protection of sensitive data with worker training. Giving employees access to tools they’ll actually use and keeping them in the loop on AI plans could help prevent the use of unauthorized tools and data leaks, foster trust, and deter sabotage of genAI initiatives.