Your Team Is Lying To You About AI. It's Not Their Fault.

Your Team Is Lying To You About AI. It's Not Their Fault.

HERE'S THE NUMBER THAT SHOULD SCARE YOU: up to 40% of employees regularly hide real concerns from their managers. Not because they're dishonest. Because they've learned that speaking up doesn't change anything, or it makes them look bad. Now put that same dynamic inside an AI rollout, where people are quietly wondering if the thing you just bought is going to replace them. You've got a room full of nodding heads and a hallway full of doubt.

I've run rollouts. I've sat in the meeting where everyone says "great, got it, makes sense" and then watched half the team quietly go back to doing it the old way the second I left the room. That's not resistance. That's self-protection. And if you're an SMB owner rolling out AI right now, you need to know the difference, because one of them you can manage and the other one you'll never see coming.

The Meeting Is Not Telling You The Truth

Group settings are the worst possible place to find out what your team really thinks about AI. Workplace silence research puts it plainly: a meaningful chunk of employees withhold real concerns from managers because they don't believe it's safe, or they don't believe it'll matter ^[1]. Add AI into the mix and you've got an extra layer of fear stacked on top, because now the thing being discussed might be connected to whether their job still exists in a year.

A recent survey found that 30% of employees admit they've faked using AI tools just to look like they're keeping up ^[2]. Read that again. Nearly a third of people are performing competence they don't have, in a system built specifically to help them, because the alternative feels riskier than the lie.

And it's not just performance around competence. It's performance around honesty itself. New research from Atlassian found that being upfront about using AI at work can actually backfire, with employees who disclose their AI use sometimes viewed as less capable or less trustworthy than those who stay quiet about it ^[3]. So now you've built an incentive structure where the smart move is to use the tool and never mention it. That's the trust gap. It's not that people don't want AI to work. It's that telling you the truth about how they're using it, or not using it, feels like it costs them something.

Leadership Believes One Thing, Employees Believe Another

This gap isn't a vibe. It's measured. RSM's Middle Market AI Survey found a real split between how confident leadership is that AI is working and how confident the actual employees using it feel day to day ^[4]. Leadership sees the dashboard. Employees see the workaround they built at 4pm on a Tuesday because the tool didn't do what it was supposed to do.

That gap matters because it means the person signing off on the AI budget and the person actually touching the tool are working from two completely different pictures of reality.

Here's the part that should really bother you: KPMG found that a large share of workers are actively hiding their AI use from their bosses, not because they're doing something wrong, but because they're not sure how it'll be received ^[5]. Some worry it'll look like they're cutting corners. Some worry it'll look like they're not needed anymore if the AI is doing the work. Either way, you end up with a workforce that treats AI like a secret instead of a tool, and you, the owner, lose the one thing you actually need: real signal on what's working and what isn't.

Why Your Team Isn't Lying On Purpose

I want to be straight with you here, because this is where most owners get it wrong. Your team isn't trying to deceive you. They're doing what any rational person does when the incentives are stacked against honesty. If speaking up about a broken workflow gets you labeled a complainer, you stop speaking up. If admitting you don't know how to use the new tool gets you labeled behind, you fake it. This is basic human behavior, not a character flaw.

Think about the position they're in. AI adoption at small businesses jumped from 23% in 2023 to 58% by the most recent count ^[6], which means most of your employees have been asked to absorb a massive shift in tooling in a very short window, often with little formal training and even less reassurance about what it means for their role. Nobody sat them down and said "this isn't about replacing you, it's about giving you leverage." So they filled that silence with worst-case thinking, and then they hid it, because admitting fear at work rarely goes well.

• They nod in the meeting because disagreement feels risky.
• They say the training was helpful because saying otherwise feels like failure.
• They quietly stop using the tool because the AI-generated output didn't fit and nobody asked why.
• They hide the fact they're using AI at all, because disclosure sometimes reads as a red flag instead of a green one ^[3].

None of that is malicious. All of it is fixable. But only if you stop trusting the meeting and start trusting the behavior.

Metrics That Lie, and Metrics That Don't

Here's the mistake I made early on, and I'll own it: I tracked the wrong things. Training completion. Logins. Meeting attendance. Every one of those numbers looked great. And none of them told me whether people actually trusted the tool enough to use it when nobody was watching.

That's the real diagnostic. Not "did they show up," but "did they change how they work when it wasn't required."

The research backs this up in a way that surprised even me. A recent workplace study found that fear around AI is actively distorting how people behave at work, changing what they say, what they try, and what they admit to, often in ways that have nothing to do with the tool's actual capability ^[7]. That means the surface-level compliance metrics you're probably tracking right now, the stuff that looks good in a board deck, are exactly the metrics most likely to mask the real problem.

So what actually tells you the truth? Voluntary use. Are people opening the tool without being told to? Unprompted criticism. Are people telling you, unasked, what's broken? Workflow change. Has the actual process shifted, or does everyone still do it the old way and just check a box that says they used AI?

The Real Cost of Getting This Wrong

This isn't a soft-skills problem you can wave off. There's real money on the table. Guidance for SMBs running AI projects suggests budgeting somewhere between 30% and 50% of total project effort and cost for change management and process redesign, not the tool itself ^[6]. That number should stop you cold if you've been budgeting for software licenses and calling it done. The tool is the cheap part. Getting your team to actually trust and use it is the expensive part, and it's the part most owners skip.

Skip it, and you get exactly what the data describes: rising adoption numbers next to rising anxiety, a team that says the right things in meetings and does something else entirely at their desk.

I think about this the same way I think about every business problem: is there a real issue here, and can I actually solve it? The issue isn't that your employees don't want AI to work. It's that you've never given them a safe way to tell you it isn't working yet. That's on leadership to fix, not on the team to somehow overcome on their own.

Your team isn't resisting AI. They're protecting themselves from a conversation nobody's made safe to have yet.

How To Actually Find Out What's Happening

Stop asking in the group setting. Start asking one person at a time, in private, with a specific question instead of a vague one. Don't ask "how's the AI rollout going." Ask "show me the last thing you used it for." Watch what they do, not what they say.

1. Track edits, rejections, and reverts, not just usage counts. If people are constantly correcting or ignoring AI output, that's your real signal.
2. Run one-on-ones instead of team check-ins for anything related to trust or confidence in the tool. People tell the truth in private far more than they do in a group.
3. Name the fear directly. If you suspect people are worried about job security, say so out loud. Silence around the fear doesn't make it go away, it just makes it go underground.
4. Reward disclosure. If someone tells you the tool isn't working for their use case, that's a gift, not a complaint. Treat it that way publicly, so the next person feels safe doing the same.

This isn't complicated, but it is uncomfortable. You have to actually want to hear that something isn't working. Most owners say they do. Fewer actually build the structure that makes it safe.

Frequently Asked Questions

Why do employees pretend to like AI tools they don't actually use?
Because the workplace incentives usually punish honesty more than they reward it. Nearly a third of employees admit to faking AI use just to appear competent ^[2], and disclosing real AI use can sometimes make someone look less trustworthy rather than more capable ^[3]. Faking it feels safer than admitting confusion or doubt.

How can a small business owner tell if AI adoption is real or just performed?
Don't trust training completion or login counts. Track whether people use the tool without being told to, whether they bring you unprompted problems, and whether their actual day-to-day workflow has changed. Those three signals are far harder to fake than a meeting nod.

Is employee resistance to AI mostly about job security fears?
It's a major piece of it. Employees frequently withhold real concerns from managers during change initiatives because they don't feel safe raising them ^[1], and that silence often traces back to fear about what the tool means for their role, not the tool itself.

How much should a small business budget for AI change management, not just tools?
A reasonable range from SMB-focused guidance is 30% to 50% of total project effort and cost going toward change management and process redesign, not the software itself ^[6]. Most owners under-budget this and wonder later why adoption stalled.

Where This Goes From Here

I've built and rebuilt businesses around exactly this kind of gap between what people say and what's actually true. It's not a comfortable place to operate from, but it's the only place where real progress happens. If you're running an AI rollout right now and you have a nagging feeling the numbers look better than the reality, trust that feeling. Go have the one-on-one conversation instead of the team meeting. Ask the specific question instead of the vague one. You'll be surprised what you hear, and honestly, that's the whole point. Come along with me as I keep figuring this out in real time, because I'd rather show you the messy version than sell you the polished one.

Sources

1. metaintro.com (metaintro.com)
2. Why employees fake AI adoption, and how you can win them over | Human Resources Director (hcamag.com)
3. New research shows honesty about AI use at work is backfiring - Inside Atlassian (atlassian.com)
4. The AI gap between leadership belief and employee confidence | RSM Middle Market AI Survey 2026 | RSM US (rsmus.com)
5. Workers Are Hiding AI Use From Bosses, KPMG Survey Finds - Business Insider (businessinsider.com)
6. AI Transforms Small Businesses, but Challenges Persist (businessinsider.com)
7. Study: AI fear is distorting workplace behavior | The Deep View (thedeepview.com)

Researched from 8 vetted sources · average source authority DR 75

Independently verified

The statistics above were independently corroborated against these sources:

• thestacc.com (DR 50)

#aiadoption #employeetrust #workplaceairesistance #employeeshidingaiusefrommanagers #airolloutcommunicationstrategies #whyteamsresistaiimplementation #smbaiimplementationchallenges #howtogethonestemployeefeedbackonai #whydoemployeeshideconcernsfrommanagers #aiworkplacetransparency #teamaiadoptionbarriers