More than a third of American workers say they have hidden their AI use from a manager, a colleague or a client. Managers are the most likely to do it. That is the headline from new research this week, and it should change how leaders think about AI disclosure at work.

It is tempting to read that as an honesty problem. I don’t think it is. When this many people, at every level, decide that the safest move is to keep the work quiet, they are telling us something about the system they work inside. They are not sure what counts anymore.

The work is getting done. The path is disappearing.

Key takeaways

  • A new Charter study of 5,000 U.S. workers found 35.9% have hidden their AI use, and managers (43.7%) hide it more than individual contributors (27.1%).
  • The top reasons are fear of looking less capable and fear of looking lazy. That is a sign effort still works as a proof signal in many organizations, even though AI has broken it.
  • Hidden AI use is invisible work. When leaders leave the rules vague, invisible work does not stay neutral. It turns into suspicion.
  • The fix is not surveillance. It is a simple, clearly communicated disclosure standard, plus managers who ask how decisions were made.

What’s happening with AI disclosure at work

Three pieces published in the past week point to the same pattern.

First, Bill Murphy Jr. reported in Inc. on October 24 on a new study of 5,000 American workers by Charter, a research and media organization focused on the future of work. Overall, 35.9% said they had hidden their AI use from a manager, colleague or client. By role, 43.7% of managers, 37.1% of executives and 27.1% of individual contributors said they had done so. Asked why, 49.3% said they worried about appearing less capable or competent, 35.5% worried about looking lazy, and 26% did not know whether their organization allowed AI use at all. The study also found that workers who said AI had increased their burnout were 28.5 percentage points more likely to have concealed their AI use than those who said it had reduced their burnout.

Second, Workday’s October AI@Work pulse survey of nearly 6,000 workers globally, published October 22, found that 41% of AI users say they have deliberately hidden their AI use on a work task, down from 49% in August. Only 35% said their employer’s disclosure expectations were very or extremely clear, and 19% said their employer had no disclosure expectations at all. The same survey found a wide gap in trust: 32% of individual contributors trust their employer’s approach to AI, compared with 66% of senior and executive leaders.

Third, in a Fast Company Executive Board column published October 27, Avenue M Group CEO Sheri Jacobs asked a simple question: does your organization have an AI disclosure rule? In her informal poll of 124 executives, 41% said employees should not have to disclose AI use because it is just a tool, 31% said always, 18% said only for external work and 10% said only if asked. When she followed up with 10 leaders, four had a formal policy, two were drafting one and four had no rule. Her conclusion: when leaders give no guidance, employees decide for themselves.

Put those together and the story is not that people are sneaking around. It is that the rules are unclear, the signals are mixed, and people are filling the gap on their own.

What this reveals: the product without the path

In Chapter 9 of The Next Turn, “Invisible Work,” I describe how work used to leave a trail. Drafts, revisions, hours, visible struggle. Leaders could watch the work happen and form a judgment along the way. AI is fading that trail.

Leaders increasingly see the product without seeing the path.

The Next Turn

Hidden AI use is the sharpest version of this. The leader gets the finished report, the analysis, the email. They do not know which parts were drafted by a tool, which parts were checked, and which parts reflect someone’s real judgment. And the person who did the work has decided it is safer not to say.

The reasons people give matter. Nearly half worried about looking less capable. More than a third worried about looking lazy. Those are not fears about AI. They are fears about effort.

In Chapter 7, “The Collapse of Effort,” I argue that AI has broken the old relationship between visible effort and visible value. For a long time, effort was how we proved we were doing good work. When AI does part of the work, people know the old proof system will read them as lazy, even when the result is better. So they hide the tool and keep the appearance. That is the collapse of effort as a signal, showing up in real time.

This is a structure of work issue. By structure of work, I mean how work actually flows: the tools, the steps, who does what. That has changed. But the definition of work, what people believe good work looks like and what earns credit, has not been updated to match. That is the collision the book is about.

And look at who is hiding it most. Managers. The people we rely on to set norms are often the least sure what the norms are. That is not a character flaw. It is what happens when leadership approves the tools but never defines the standard.

If those definitions remain vague, invisible work does not stay neutral. It turns into suspicion.

The Next Turn

Suspicion runs both directions. Leaders start to wonder whether fast work is real work. Employees start to wonder whether honesty will cost them. The Charter finding on burnout suggests that secrecy itself carries a cost for the people keeping it.

Why it matters for leaders

In Chapter 12, “Defining Work Again,” one of the seven moves is to define AI’s role in the work before employees define it informally. These surveys show what happens when that move is skipped. One in four workers in the Charter study did not even know whether AI use was allowed. Fewer than four in ten in the Workday survey said disclosure expectations were clear.

Put simply, ambiguity creates its own behavior. People hide AI use, overuse it, or distrust fast output.

The wrong response is more monitoring. Surveillance gives you more data about activity. It does not give you more clarity about judgment. The right response is to make thinking visible, even when every step of the process is not.

That is also where standards and signals come in. The standard has not changed. You still need accuracy, sound judgment and accountability. What changed is the signal. “I spent six hours on this” used to stand in for quality. It no longer can. A clear disclosure norm helps you build a new proof system: what did the tool do, what did you decide, what did you verify?

Disclosure is not a confession. It is context. And context is what leaders need to evaluate judgment.

Your next turn

  • Ask your team one question this week: “If you used AI on something and didn’t mention it, what were you worried would happen?” Listen for whether the answer is about rules or about reputation.
  • Write a disclosure standard that fits on one page. Name where AI use is expected, where it should be disclosed, where it needs human verification, and what can never be entered into an outside tool. Match the level of disclosure to the level of consequence.
  • Go first. If you use AI in your own work, say so, and say what you checked. Managers are the group most likely to hide it, so their openness carries the most weight.
  • Change one review question. Replace “How long did this take?” with “What did you trust, and what did you verify?”
  • Check your reward system. If fast work done with AI quietly gets less credit than slow work done by hand, people will hide the tool no matter what the policy says.

Frequently asked questions

Why do employees hide their AI use at work?

In the Charter study reported by Inc., the most common reasons were worry about appearing less capable (49.3%) and worry about looking lazy (35.5%). About a quarter (26%) did not know whether AI use was allowed. Those answers point to unclear rules and outdated ideas about effort, not dishonesty.

Should companies require employees to disclose AI use?

Most organizations benefit from a clear, simple disclosure standard, but it does not have to mean labeling every email. A practical approach matches disclosure to consequence: higher-stakes, external or decision-making work needs more transparency than routine drafting. What matters most is that the rule is easy to understand and communicated regularly.

How should managers evaluate work that was done with AI?

Evaluate decisions, not just deliverables. Ask what the tool produced, what the person changed, what they verified and why. The standards stay the same: quality, accuracy and sound judgment. What changes is the evidence you use to see them.

There is a pattern I keep seeing in my work with leadership teams. Organizations invest in AI tools, encourage people to use them, and then evaluate work with signals built for a world without them. People notice the gap. They respond rationally. They keep the work quiet.

As I put it in Chapter 12, “AI may participate in the work. It does not inherit responsibility for the work.” A clear disclosure norm is how you keep responsibility visible, and how you give people permission to be honest about how the work really got done. You can find more on these ideas on the Next Turn blog.

So here is the question worth sitting with: if someone on your team used AI to do their best work this week, would they feel safe telling you?

Sources