November 20, 2026

Majority of Workplace AI Use Falls Outside Enterprise Control, New Research Reveals

2151998481

More than eight in ten workplace AI interactions take place outside enterprise accounts, according to new research from CultureAI, raising concerns about how businesses are managing AI adoption and protecting sensitive information.

The AI security and governance company’s latest report, The Unmanaged Majority, analysed nearly 1.7 million AI interactions across 40 organisations during Q2 and Q3 2026. It found that 81% of AI activity took place through personal, free or unidentified accounts, potentially leaving businesses without the visibility and safeguards offered by enterprise agreements.

The research also revealed that sensitive information was detected in approximately one in every seven AI prompts, highlighting the risks associated with employees increasingly incorporating AI tools into their everyday work.

Businesses Struggling to Keep Pace with AI Adoption

The study identified 1,064 different AI applications being used across the organisations analysed, with an average of 113 applications per business. Nearly half of these tools were found in just one organisation, suggesting that employees are experimenting with a much wider variety of AI applications than businesses may realise.

Interestingly, general-purpose chatbots accounted for just 6% of the applications identified, demonstrating that workplace AI adoption extends far beyond familiar tools such as ChatGPT.

Despite businesses increasingly investing in approved AI platforms and enterprise licences, only 19% of interactions took place through enterprise accounts.

A further 47% involved personal, free or unauthenticated access, while 34% occurred through applications where the licence type could not be determined.

These findings suggest that traditional approaches to AI governance, which often focus on approving specific applications and introducing usage policies, may be overlooking the majority of employee activity.

Blocking AI Tools Could Make the Problem Worse

The research also found that restricting access to AI applications does not necessarily prevent employees from using them.

In organisations that blocked AI tools, personal account usage increased by approximately 50% following the restrictions.

This suggests that employees may be finding alternative ways to access the technology, potentially moving their activity further beyond the visibility of corporate IT and security teams.

As businesses look to capitalise on the productivity benefits of AI, the findings highlight the importance of balancing security requirements with employees’ growing demand for these tools.

Sensitive Business Information Regularly Shared with AI

Alongside concerns about visibility, CultureAI’s research uncovered significant levels of sensitive information being shared with AI applications.

Across the interactions analysed, researchers recorded 252,480 sensitive data detections spanning 100 categories.

HR records and correspondence were identified in 90% of organisations, while company strategy information appeared in 85%. Health-related information was detected in 82% of organisations, with legal case details and financial information each identified in 72%.

Perhaps surprisingly, credentials accounted for less than 1% of all sensitive data detections. Instead, information relating to company strategy, finance, HR, legal matters and health outnumbered credential exposures by almost 30 to one.

This presents a challenge for businesses relying on traditional data protection technologies, which are often designed to recognise structured information such as passwords and financial account numbers rather than contextual information contained in everyday workplace documents.

The findings also carry potential regulatory implications. Depending on the circumstances, sharing personal or confidential information with external AI platforms could raise concerns under UK GDPR and other data protection requirements.

Businesses Urged to Rethink AI Governance

Oliver Simonnet, Lead Cybersecurity Researcher at CultureAI, said the findings demonstrate a growing disconnect between organisations’ expectations and the reality of workplace AI use.

“AI is now firmly embedded in everyday working life, but our research shows a significant disconnect between how organisations believe AI is being used and what is actually happening. With 81% of AI interactions taking place outside enterprise accounts, it’s clear that traditional approaches to governance are failing to keep pace with employee behaviour.”

Simonnet added that businesses need to move beyond simply approving AI tools and introducing policies if they want to manage the risks effectively.

“Organisations cannot assume that approving an AI tool or introducing a policy automatically means they have control over how it is used. They need visibility into the applications employees are accessing, the accounts they are using and, crucially, the information being shared.”

He also cautioned against relying on blanket restrictions, arguing that organisations should instead focus on enabling employees to use AI securely.

“Restricting AI isn’t the answer. Where organisations blocked AI tools, personal account use rose by around half. The better route is to make secure use the easiest option. By continuously monitoring AI activity and applying appropriate safeguards, businesses can protect sensitive information without sacrificing the productivity benefits that make AI so valuable in the first place.”

As AI becomes an increasingly established part of everyday business operations, the research suggests that organisations will need to rethink how they approach governance, moving towards greater visibility of employee activity and more effective safeguards for sensitive information.

Image source: Magnific