WalkMe's 2026 Adoption Report: 54% of Your Team Is Bypassing the AI You Mandated

Eighty percent of white-collar workers are actively avoiding the AI tools their employers paid to deploy.

That number comes from WalkMe’s fifth annual State of Digital Adoption report, published this week: a survey of 3,750 executives and employees across 14 countries. Fifty-four percent bypassed their company’s AI tools in the past 30 days and did the work manually. Another 33% haven’t touched the tools at all.

These are not companies that skipped training. These are organizations that have deployed, trained, and in many cases mandated AI use. The tools are sitting there. People are choosing not to use them.

The conventional response to that data is to order more training. I want to make the case that this is exactly the wrong call.

The myth

When AI adoption stalls, the instinctive response is more training. Bigger onboarding programs. Mandatory workshops. Lunch-and-learns about prompting. The logic holds together: people aren’t using the tools because they don’t know how. Get them trained, adoption follows.

This is the most expensive, least effective intervention available to most leadership teams right now. Not because training is useless, but because it’s solving the wrong problem.

Why it sounds right

The skills gap is real, and that makes the diagnosis easy to reach. Fifty-six percent of workers globally report receiving no recent AI skills development. Adoption rates vary by department, by seniority, by whether someone happened to sit next to a power user during a demo. When you watch people struggle with the tools, the conclusion writes itself.

Training is also a comfortable solution. It’s measurable, procurable, and communicable. “We invested in our people” is a clean board narrative. It lets the organization respond to adoption problems without disrupting how the work actually gets organized, evaluated, or rewarded.

That comfort is exactly why it keeps getting chosen even when it doesn’t work.

What the evidence says

The WalkMe report has one number I keep coming back to: only 9% of workers trust AI for complex, business-critical decisions. The corresponding figure for executives in the same survey was 61%.

That is a 52-point gap. And it exists inside companies that have already deployed the tools, already run the training, already communicated the mandate.

A training program cannot close a trust gap. Trust in AI at work is not about whether someone knows how to write a prompt. It’s about whether the organization has given them reason to believe that using AI well will be good for them specifically: for their standing, their workload, their career trajectory. Without that foundation, the skills are beside the point.

The second piece of data came out of a Writer and Workplace Intelligence survey published in Fortune on April 8. Two thousand four hundred knowledge workers across the U.S., U.K., and Europe. Twenty-nine percent of employees admitted to actively sabotaging their company’s AI strategy. The figure among Gen Z workers was 44%. The primary motivation, cited by 30%, was fear that AI would eliminate their jobs.

The irony here is worth sitting with. The people most afraid of being displaced by AI are the ones making themselves most vulnerable to displacement by avoiding it. But their fear is not irrational. What they are missing is not skills. It is clarity: on what happens to their role, their status, their job security if they become genuinely proficient. That question does not get answered in a training session.

There is a third data point, from Gartner research also released this week: only 7% of organizations guide employees on how to use the time that AI frees up. So companies train people to work faster, then leave them in a void about what “faster” means for their position. That is a work design failure, not a skills failure. And the result is predictable: people who see increased capacity as a threat to their headcount justification will find ways to fill that capacity with manual work.

The tool adequacy numbers from WalkMe are just as striking. Eighty-eight percent of executives believe employees have adequate tools. Twenty-one percent of workers agree. That is a 67-point gap on something as basic as “does the tool work for my actual job?” The executives buying the platforms are not the ones using them in context. What reads as adequate in a demo often doesn’t survive contact with real workflows.

The reframe

AI adoption resistance is not a capability failure. It is a trust signal, and it is specific.

Workers are not saying they cannot use AI. They are saying they do not believe that using AI is safe for them. Safe for their career. Safe for how their performance is evaluated. Safe given that nobody has told them what the organization will do with the productivity gains.

The companies seeing real adoption have not necessarily run better training programs. They have done three things differently. They have named explicitly what AI proficiency means for career trajectory. They have put managers, not just trainers, in the role of showing teams how and why to change how they work. And they have answered, publicly and specifically, the question most employees are too uncertain to ask aloud.

Training teaches the what. Leadership has to answer the why. Most organizations are delivering the former while hoping the latter takes care of itself.

The Gartner finding is useful here: 46% of managers are experimenting with AI in their own work, compared to 26% of employees. The gap is not between the executive suite and the floor. It runs through the manager layer, which is exactly where adoption decisions get made day to day.

So what

Before ordering the next training cohort, run the diagnostic first. Not “do they know how to use the tools?” but “do they believe that using the tools is good for them?”

If the answer to the second question is no, more training will not move the number. It might even make it worse, because it signals that leadership has diagnosed the wrong problem.

The executives I have seen actually shift adoption rates have done something different from anything in a standard change management playbook. They have made the implicit explicit: here is what we expect proficiency to look like, here is what happens to the people who get there, and here is what we are doing with the time that gets freed up. Not as a policy memo. As a visible leadership behavior repeated until people believe it.

Training is part of the picture. It is not the picture.