A recent article on CIO.com made a sharp observation that deserves to be taken further. The author’s core argument: Organizations are reporting AI activity to their boards — tools purchased, pilots launched, licenses deployed — while quietly avoiding the harder question of whether any of it has actually moved the business. Outcomes were never defined before the projects began, so success cannot honestly be measured after the fact. The board hears momentum. The CFO sees cost. And nobody can clearly answer what actually changed because of AI.
It is a well-observed problem. But it only tells half the story.
The other half is happening desk by desk, in organizations everywhere. While executives debate ROI frameworks, a parallel economy of AI productivity is running quietly in the background — driven by employees who have figured out how to use these tools and have calculated, quite rationally, that the safest thing to do is say nothing about it.
Understanding what is driving that silence is not a secondary concern. It is arguably the most important AI management challenge most organizations have not yet named.
The job security calculation no one talks about
The most important driver of AI silence is also the most understandable.
Consider an employee who has quietly been using an AI tool to draft client reports. A task that once took four hours now takes 45 minutes. The output is better: Tighter, better structured, more thoroughly referenced. Her manager is pleased. Her clients are happier. And she has said absolutely nothing to anyone about how she is doing it.
When employees find themselves in this situation, the reasoning for staying silent is almost always the same: If I tell them I can do it in 45 minutes, they’ll wonder what I’m doing with the rest of my time. Or they’ll give me more work. Or they’ll decide they don’t need as many of us.
This is not paranoia. The ADP Research Today at Work 2026 report — which surveyed more than 39,000 workers across 36 markets — found that only 22% of global workers strongly agreed their job was safe from elimination, even against a backdrop of historically low unemployment. The culprit identified by the report is AI anxiety, gripping workforces regardless of seniority or sector.
The scale of that anxiety has a structural basis. The World Economic Forum’s Future of Jobs Report 2025 (weforum.org) found that while 77% of employers plan to upskill staff to work alongside AI, 41% simultaneously plan to reduce their workforce as AI automates certain tasks. Employees are reading those numbers carefully, even when their employers are not.
Research from Ivanti puts the scale of the resulting silence in sharp relief: Nearly one-third of workers keep their AI use secret from their employer, with 30% specifically citing fear that their job will be cut if they disclose it, and a further 36% staying silent because they enjoy the competitive edge AI gives them over peers. The employee who is most proficient with AI — and therefore delivering the greatest productivity uplift — has the most to lose by saying so. So, they say nothing, the gain disappears invisibly into expanded workload, and it never surfaces in any report to the board.
For organizations trying to understand the true impact of AI on their operations, this is a foundational measurement problem. The biggest wins may be the ones most deliberately hidden.
What’s actually happening beneath the surface
The job security calculation is the most significant driver of AI silence, but it is not the only one. At least four other dynamics are keeping the real story from reaching leadership:
- Competitive concealment, which the Ivanti data captures well. Not every employee who hides AI use is afraid of their employer — some are protecting an edge over colleagues. In performance-ranked environments — sales floors, bid teams, content departments — knowing how to use AI effectively is increasingly the difference between hitting targets and missing them. People who have that edge are not always eager to share it.
- What the data describes as complacent and non-transparent use. A KPMG global study of more than 48,000 workers across 47 countries found that 58% of employees are intentionally using AI at work, yet the study identified widespread non-transparency in how it is being used — with many not checking the accuracy of AI outputs or disclosing usage to managers. Nicole Gillespie, co-author of the report and a professor at the University of Melbourne, described the findings as a troubling level of “inappropriate, complex and non-transparent” AI use. Her prescription: Organizations must create transparent, shared learning environments where employees feel safe to experiment with AI without fear.
- The third is something harder to name: A kind of impostor anxiety. The same Ivanti research found that 27% of employees who use AI at work experience impostor syndrome as a result — they feel that the quality of their AI-assisted work is better than what they could produce alone, and that this gap is somehow dishonest. These tend to be the most thoughtful and quality-conscious adopters in the organization, and they are actively obscuring the AI contribution to their work rather than risk being seen as relying on a crutch.
- The shadow infrastructure problem. A Laserfiche-commissioned survey published in Security Magazine (securitymagazine.com, August 2025) found that 49% of American employees hide their AI tool use from their employer, with only 36% reporting clear AI guidelines and an approved tools list in their workplace — and one in ten describing their organization’s AI environment as “the Wild West.”
The data security implications run deeper still. The KPMG global study found that 46% of US employees have uploaded sensitive company data into public AI tools, often without knowing whether the content was confidential. That is not malicious intent — it is the predictable result of a governance vacuum — but it represents a risk exposure that leadership is largely unaware of.
What leaders should actually do about this
These four dynamics — fear of redundancy, competitive concealment, impostor anxiety and shadow infrastructure — combine to produce a fifth and arguably most damaging outcome: The “do more with less” spiral.
When employees quietly use AI to work faster, organizations rarely recognize the efficiency gain and redistribute the capacity thoughtfully. They simply load those employees with more work. The report that used to take four hours now takes 45 minutes, so more reports get assigned. The workload expands to absorb the freed capacity. The employee cannot now reveal the AI assistance without exposing how much time they have been quietly banking. And so, the spiral continues: More output, more concealment and no organizational learning captured.
The CIO.com article’s central argument — that organizations must define outcomes before embarking on AI projects — is correct. But the hidden dynamics described above suggest the measurement problem runs deeper than an absence of pre-defined success criteria. You cannot define meaningful outcomes if the people generating the most significant AI-driven results are structurally incentivized not to tell you about them.
Closing that gap requires organizations to make three interconnected shifts — each designed to tie AI’s business outcomes directly to the employees doing the work and to create the conditions in which those employees are willing to share what they know.
Step 1: Make the commitment explicit — and tie it to outcomes from the start
The first step is to make an unambiguous public commitment that AI productivity gains will not be used as the basis for headcount decisions. But a commitment alone is not enough — it only holds weight when it is paired with something concrete employees can see: Business outcomes defined before the project begins, not after.
Not “AI will make us more efficient” — which means nothing and measures nothing — but observable, agreed results: Client proposal turnaround reduced from five days to two; compliance review time cut by 40%; customer query resolution improved by a defined margin within a defined period. A CIO.com analysis of AI metrics found that the most effective organizations evaluate success across three dimensions: Return on employees (output per hour, backlog reduction), return on investment (labor cost per worker, conversion rates) and return on future (market share signals, new capability unlocked). None of those measures require employees to justify their existence. All of them create a shared definition of what winning looks like.
When business outcomes are defined upfront, the dynamic shifts. Employees can see that the measure of AI’s success is the outcome — not their hours logged or headcount consumed. Leadership has something meaningful to report to the board beyond adoption figures. And the question changes from “how many people are using AI?” to “what did AI change about this result?” — a question employees can answer honestly, because the answer no longer puts their role at risk.
Step 2: Build incentives strong enough to override the fear
This is the step most organizations skip entirely — and it is the most important one. A commitment not to cut jobs and a clear outcome framework create the conditions for honesty. But it does not actively reward it. For employees who have spent months quietly banking efficiency gains, the rational calculation remains: Why surface what I have if there is nothing in it for me?
The answer from the organizations doing this well is: Make sharing genuinely worth it. Not as a vague cultural aspiration, but as a structured, visible program with real rewards attached.
Wharton senior fellow Scott Snyder has proposed treating employee time as capital: If an individual identifies an AI method that saves four hours a week, they receive a portion of that saved time — perhaps 50 hours a year — to invest in further AI experimentation or professional development. This creates a direct incentive to disclose efficiency gains rather than conceal them, and it transforms the calculation from “what do I lose by sharing?” to “what do I gain?”
Real-world examples are already emerging. Law firm Shoosmiths created a £1 million bonus fund tied to Microsoft Copilot usage, with 1,300 employees eligible to receive approximately £770 each if the firm reached one million Copilot uses in its fiscal year. IBM awards “BluePoints” to winners of its annual AI innovation contest, redeemable for electronics, appliances or event tickets. Pharma firm Sanofi uses a points system to reward employees who experiment with AI and share what they learn. As Sanofi’s head of culture put it: “Recognition is the fuel of trust, and trust is what makes AI adoption possible and scalable.”
McKinsey’s 2025 workplace AI research confirms that 40% of employees say incentives and financial rewards would increase their daily use of AI — ranking it fourth among the factors that would most improve adoption, behind training, workflow integration and tool access. EY’s Work Reimagined survey goes further, finding that organizations that formally align rewards with AI behaviors and outcomes are significantly more likely to achieve transformational results, while those that deploy AI onto “fragile talent foundations — weak culture, insufficient learning, misaligned rewards” see productivity benefits lag by over 40%.
The principle behind all of these approaches is the same: Employees will share the benefits of AI when sharing the benefits of AI is rewarded — concretely, consistently and visibly. Until that condition is met, the most productive employees in the organization will remain the quietest.
Step 3: Rebuild the board update around outcomes and employee voice
Third, demand more from the board update. Grant Thornton’s 2026 AI Impact Survey found that organizations with fully integrated AI are nearly four times more likely to report revenue growth than those still piloting. The difference is not primarily technological — it is governance and accountability. The leading organizations can demonstrate how their AI makes decisions, who owns the outcomes and what happens when something goes wrong. That level of transparency can only exist when both leadership and employees are operating in the open.
A board update built around outcomes looks fundamentally different from one built around activity. It does not lead with “We have deployed AI across fourteen workflows.” It leads with “Here is what changed in the business because of AI, here is how we measured it and here is what our employees told us about working with it.”
That last element — what employees said — is not a soft add-on. A CIO.com piece on AI adoption published in 2025 put it plainly: “Trust is the invisible infrastructure of AI adoption. It’s built through transparency about intent, honest conversations about job impact, visible upskilling opportunities and letting employees see their peers genuinely benefit.” Employee willingness to use AI, and to share its benefits openly is the most reliable leading indicator of whether an AI program is building genuine organizational capability or simply burning through budget on tools that will be quietly worked around.
Organizations that track this systematically ask three questions on a regular basis: Is AI use growing organically, or only where it is mandated? Are employees who use AI more likely to flag further opportunities, or do they stay quiet? And when AI delivers a measurable outcome, does the team responsible feel able to claim it?
If the answers are “mostly mandated,” “they stay quiet” and “not really” — the organization has a trust and incentive problem that no amount of AI investment will solve. The technology is not the constraint. The environment is.
The board update on AI should not just report how many licenses are deployed and how many pilots are underway. It should grapple with harder questions: What are employees actually using AI for today, including tools we did not procure? What outcomes has that usage produced and how do we know? What would it take to make it safe — and genuinely worthwhile — for them to tell us?
Until those questions are asked — and until the answers can be given without fear and with something to gain — the most important AI story in the building will continue to be told in silence. The board will keep hearing about activity. The CFO will keep questioning ROI. And the employee who cracked the code months ago will keep her head down, produce excellent work and say nothing.
That is the measurement problem the CIO.com article did not quite reach. And it is the one that matters most.
This article is published as part of the Foundry Expert Contributor Network.
Want to join?