Every organization is investing money in AI right now, and almost none of them know what their people are actually doing with it": Study reveals employees are using their personal AI accounts at work, raising a host of problems

Two-thirds of AI use in personal accounts is actually for work purposes
Workers are also using company-provided tools to ask personal questions.
Clumsy enterprise authentication makes it harder to access approved tools in an instant

New research from Harmonic has claimed that almost two-thirds (64.5%) of all activity on personal and free AI accounts is actually for work purposes, meaning there is a significant amount of AI usage going completely unnoticed by businesses.

At the same time, enterprise-level accounts are used for personal questions, meaning employees and AI meet wherever convenient, regardless of security policies. In fact, nearly half (45.6%) of all personal AI activity occurs on licensed plans paid for by companies.

In reality, workers don’t treat work AI and personal AI as separate things, but instead bring their tasks to whatever AI tool is already open or easily accessible on their device, regardless of whether it’s employer-provided or personal, free or paid.

Undetected work on personal AI is creating a visibility gap

Harmonic’s research serves to highlight the visibility gap that is emerging as AI adoption spreads, with legal and governance workers having the greatest use and visibility. These workers account for about a fifth (19.5%) of all AI hours across all teams, and 81% of this usage occurs with approved tools.

Marketing teams are the second largest users, at 17.5%, but only 39% of GTM AI activity occurs in enterprise-approved tools, resulting in poor visibility. But that’s still twice the visibility of operations teams, where not even a fifth (18%) of activity is executed on business plans.

As for why AI is used at work, the clearest purpose is efficiency and automation (47%), which ranks well ahead of decision support (20%) and risk and compliance (20%). Revenue and growth (7%) and innovation (6%) are less common.

The true measure of usage is minutes, not queries.

Where Harmonic’s research differs from other studies is in its use of “minutes” rather than “total visits,” which it says provides a much truer reflection of usage patterns. Longer sessions indicate more data exposure, he says, and Claude comes in first for actual minutes (10 min 12 sec) compared to ChatGPT (5 min 53 sec).

This is especially problematic when workers choose to use their own personal AI accounts, because sensitive company information and business context remain in their personal AI history even when they leave the company. Organizations do not even have the legal or technical powers to delete or recover that data, leading to permanent loss of intellectual property.

The path of least resistance

Harmonic explained that many companies implement strict and complicated authentication processes for enterprise AI tools, making personal tools much easier to use. Popular personal tools like ChatGPT, Gemini, Claude, and Perplexity also require little more than a Google (or similar) account to log in.

All this while companies pay a premium for licenses that are barely used: Microsoft 365 Copilot is commonly deployed at $30 per user per month; ChatGPT Business plans cost between $20 and $25 per month.

“Every organization is pouring money into AI right now, and almost none of them know what their people are actually doing with it,” summarized Alastair Paterson, CEO of Harmonic Security, noting that this is the first study of its kind to uncover how AI is “actually used at work.”

Clearly, the problem is not necessarily the provision of incorrect tools, but rather ease of access. Looking ahead, businesses are encouraged to adopt universal single sign-on (SSO) to make login easier. But Harmonic still questions the “one size fits all” approach, urging employers to consider workflows and provide the right tools to the right teams.

Google logo on black background next to text that says