Your team uses AI and isn't telling you: how to make it safe to say so
Picture your best analyst delivering in half the time. You ask how, and the answer is "I just organize myself well." Maybe that's true. Or maybe they use an AI tool nobody else knows about, and they don't say so because they don't know what would happen if they did.
Key takeaways
- Two studies (USC, 604 employees; Istanbul, 635) link hiding AI use to fear and low trust
- Nearly 1 in 3 in the USC study withheld AI techniques from coworkers or employers
- Surveys: associations, not causes, and no Mexico data
- Our recommendation: react with curiosity, make the benefit visible, keep the data rule separate
Picture your best analyst delivering in half the time. You ask how, and the answer is "I just organize myself well." Maybe that's true. Or maybe they use an AI tool nobody else knows about, and they don't say so because they don't know what would happen if they did. Two recent studies suggest this is common, and that the cause has less to do with rules and more to do with trust.
What the studies found
University of Southern California (USC) study
Researchers Eric Anicich and Jeslyn Brouwers at USC Marshall surveyed 604 US employees who say they use AI daily or several times a day. According to the university's write-up:
- Four out of five believe sharing their techniques would benefit their team.
- Nearly one in three admitted withholding AI-related knowledge, workflows or techniques from coworkers or employers.
- People in the bottom quarter on trust in their organization were about four times as likely to have withheld something as people in the top quarter.
The authors describe three fears: that admitting AI use will hurt their reputation, that the boss will treat the gain as a reason to pile on work (or make them the company's "AI spokesperson"), and that showing what the tool can do will reveal it could replace them.
Marketing study from Istanbul
A paper published in June 2026 in Behavioral Sciences surveyed 635 marketing employees who already used AI, twice. It calls hiding AI use after the fact "AI disclosure silence." Its abstract says no single condition was a strong necessary one: the silence showed up through several pathways, including AI anxiety, fear of negative evaluation, feeling creativity is threatened, job insecurity, low trust in management, weak psychological safety and unclear AI policy. The authors say the findings are configurational and predictive, not causal, and call for validation across sectors and cultures.
How to read this
These are self-reported surveys: they show associations, not what causes what. The Istanbul study covers marketing staff in one city; the USC study covers employees who already use AI often. We don't know whether they apply to your team, and we found no Mexico data.
Why it matters to an owner
If your people hide how they work, you decide with the wrong picture: you don't know which tools touch customer data, which method explains someone's good results, or how much of what you see a person did and how much a machine did. The USC researcher puts it this way: if your best analyst is fast because of a method nobody has seen, you can't train anyone else on it, and you don't really know what you're paying for.
The next question is how much authority to give those tools: how much authority should you give Copilot Autopilot in your business.
What you could do (our recommendation)
This mixes what the USC authors say with our own judgment. We label it: it is not proven.
How to tell it's working
- People start telling you about tools you didn't know.
- Questions about what can be pasted go up, not down.
- Someone adopts a method another person showed.
Why a policy alone is not enough
The USC researchers say it plainly: releasing a policy that claims to support AI use is not enough. Sending a memo saying we are now a culture that shares innovations is different from showing it. The Istanbul study points the same way: policy clarity and trust in management were associated with less silence, but more weakly than fear of being judged and the sense that creativity was threatened. That is a survey association, not proof, but it suggests another written rule does not replace daily behavior.
And the security of those tools needs a clear owner: who owns security when AI starts taking actions.
Questions to ask yourself this week
- The last time someone told me they used AI, what did I do with that information?
- Does my team know which data never goes into a tool, or are they guessing?
- Who comes out ahead in my business when someone finds a faster way to work: that person, or only me?
A sentence to start with
"I know many of you already use AI and that's fine with me. I'm not trying to police anyone. I want to know what works so we all benefit. The only thing I ask is that you don't paste customer data into tools we haven't reviewed." (Sample wording: adjust it to your style.)
Limits
Two studies don't make a rule. Both measure attitudes, not business results. Our list hasn't been tested with real teams. If your business is very small, a direct conversation may be enough.
Bottom line
The available evidence points to hiding AI use being more tied to fear and trust than to unclear rules. Before writing another policy, check how you react when someone tells you they use it.
Sources
- USC / Phys.org, "Why are employees reluctant to disclose AI use to their bosses?" (Sep 28, 2026): phys.org
- Mizrak and Karakaya, Behavioral Sciences 16(6):994 (Jun 15, 2026), abstract on PubMed: pubmed.ncbi.nlm.nih.gov
- Trestlewire (Oct 4, 2026), secondary coverage of the Istanbul study (we did not use its figures from other surveys): trestlewire.com
Was this article helpful?