← Insights · August 30, 2026

Your Team Is Hiding Their AI Use From You. A Policy Won't Fix That. You Will.

Picture the scene, because it's already happening on a job you run. A PM pulls up a chatbot to draft a tricky subcontractor email, gets it most of the way there, and finishes it by hand. Nothing wrong with that. Except when your footsteps hit the trailer stairs, the tab closes first. Not because there's a rule against it. There probably isn't. He closes it because some part of him has decided that letting you see it costs him something, and he'd rather not find out what.

That instinct is the whole story, and it turns out it's not rare. Harvard Business Review's research on AI transparency in the workplace, published in June 2026, found that employees develop genuinely valuable AI workflows through private experimentation and then choose not to share what they've learned, not mainly because of weak tools or unclear governance, but because they don't trust what their organization will do with that knowledge once it becomes visible (Harvard Business Review, June 10 2026). The Star's reporting on a separate 2026 survey put a number on the same pattern: 52% of workers hesitate to admit using AI on their most important tasks, and 53% worry that admitting it makes them look replaceable (The Star, August 21 2026). Cybernews found something adjacent from a different angle entirely, a security-focused survey rather than a trust one: 59% of employees admit to using AI tools their employer never approved, often specifically because the sanctioned tools don't do what they need (Cybernews, August 2026). Three different studies, three different lenses, one consistent finding underneath all of them: your people are already using AI, and a meaningful share of them have decided you're the last one who should know.

The concealment isn't a policy gap. It's a trust reading.

The instinct when you hear "employees are hiding AI use" is to reach for a policy fix: write a clear AI-use guideline, put it in the handbook, problem solved. HBR's research is specifically pointing away from that instinct. Employees weren't confused about whether AI use was allowed. They were making a calculated read of whether disclosing it was safe, and landing on no. That's not a documentation problem. That's a trust problem, and trust problems don't get solved by better paperwork.

Here's where it gets uncomfortable for anyone in a leadership seat: HBR's research also found that leaders teach this concealment, usually without meaning to, through small social signals. A joke about someone "cheating" with AI. A compliment aimed at the guy who did his takeoff "the old-fashioned way," meant as praise for effort, landing as a verdict on anyone using the tool. None of that is a policy. It's tone, and tone is exactly what a foreman or a PM is reading when he decides whether to close the tab before you walk in.

What actually moves the needle isn't a mandate. It's you, visibly using it.

If concealment is the symptom, the obvious next question is what actually fixes it. And this is where the second piece of research matters, because the answer isn't more oversight and it isn't a top-down mandate either. Microsoft's 2026 Work Trend Index, built on a mixed-methods study including a 1,800-person global survey on managers and AI, found that when managers actively modeled their own AI use in front of their team, rather than simply permitting it or requiring it, employees reported a 30-point lift in trust in AI and a 22-point lift in their own critical thinking about how they used it themselves (Microsoft's 2026 Work Trend Index). The same report tested 29 separate factors across organizational environment, individual mindset, and demographics, and found that organizational factors like culture and manager behavior account for 67% of AI's real impact, more than double the 32% attributable to individual mindset (Microsoft's 2026 Work Trend Index). A separate analysis of that same report put the gap in even starker terms: workers whose managers openly use AI are dramatically more likely to be genuinely skilled, high-frequency users themselves than workers whose managers don't (CXM.world).

Sit with what that actually means. It's not that a policy failed to be clear enough. It's that a mandate from above, "the org is adopting AI, get on board," and a leader's own visible behavior are two entirely different signals, and employees are responding to the second one, not the first. A memo tells people what's allowed. Watching your own leader actually use the thing, badly at first, out loud, in front of you, tells people what's safe.

What "visible" actually looks like on a jobsite, not a slide deck

This isn't abstract for me. I've used AI tools daily in my own management work, on my own time, with public tools, nothing employer-specific about any of it. What changed the dynamic with the people around me wasn't announcing that I used it. It was letting them actually see it happen, a draft email pulled up mid-conversation, a schedule conflict I ran through a tool in front of someone instead of behind a closed laptop, admitting out loud when the output was wrong and I had to fix it myself. That last part matters more than the rest. The moment you let your team watch you correct a bad AI output, you've told them something a policy never could: using this thing doesn't mean you stopped thinking, and neither does it mean they have to stop either.

Compare that to the more common version of "AI leadership" happening right now: a top-down rollout, a required tool, a training session, and a leader who never actually opens the thing themselves in front of anyone. That's a mandate wearing the costume of adoption. Per the research, it's also close to the least effective version available, because it skips the one variable that actually moves people: watching someone they trust use it imperfectly and stay credible anyway.

Why this is a leadership problem before it's a tooling problem

This is the pattern Vantage Ops keeps coming back to, and it's worth naming directly here: the thing that actually changes with AI isn't primarily what gets automated. It's the relationship between a leader and the people who work for them, what gets said out loud versus what gets read between the lines, what a joke in the trailer teaches a whole crew about what's safe to admit. Concealment and adoption aren't two separate problems requiring two separate fixes. They're the same relationship, read two different ways. A team that's hiding its AI use from you and a team that trusts you enough to show you are responding to the exact same set of signals you're already sending, whether you meant to send them or not.

You don't fix that with a better handbook policy, and you don't fix it by mandating a tool from the top down either. You fix it the way you'd fix any other trust gap on a crew: by being the one who goes first, visibly, imperfectly, where people can actually see it.

Your team already knows whether you're one of the leaders who jokes about AI or one who's actually willing to be seen using it. The only question left is which one you've decided to be.

This is the exact gap Vantage Ops works with leadership teams to close, not by handing over a rollout plan or a tool license, but by working through how you personally show up as AI enters your operation, since that's the part your team is actually watching.

Talk to Vantage Ops about your team