Issue 269, June 18, 2026
A senior director at a professional services client recently shared something with me that she had not told anyone else in her organization. Her team of eleven had quietly become one of the most productive units in the company. Turnaround times had dropped. Client deliverables had improved. Her leadership noticed and praised the team’s discipline. What her leadership didn’t know, and what she had decided not to tell them, was that nearly everyone on her team was using AI tools daily, none of which had been approved, sanctioned, or even discussed by the organization. When I asked her why she hadn’t disclosed it, her answer was immediate and unguarded. “Because I don’t know if we’re allowed. And if I ask, the answer might be no. And then what do I do with a team that has already changed how it works day to day?”
She is not an outlier. She is the new norm. Talk to most around you, and you are likely to hear the same or similar. The dynamic she described to me deserves a name, because it’s quietly reshaping organizations everywhere while leadership looks at dashboards that no longer describe how work actually gets done. I call it The Permission Paradox: the conditions in which an organization and its leadership have neither permitted nor prohibited a new way of working, leaving every individual to fend for themselves and resolve the question privately. Which in turn makes concealment the most rational choice available to them.
A Note on What This Newsletter Is and Is Not About
Before going further, a word about the territory we have been covering. Readers of recent issues might reasonably wonder whether Ideas and Innovations has become an AI newsletter. It hasn’t, and it won’t. This newsletter has always been about the human factor: the psychology, identity, and behavior that determine whether organizations change or merely announce that they intend to. What has changed is that AI has become the most revealing stress test of human behavior that organizations have encountered in decades. It surfaces dynamics that were always present but easy to ignore. How people respond when their competence is threatened. What they do when guidance is absent. Whom they trust, and whom they protect themselves from. AI is not the subject. It is the lens. And right now, the lens is showing us things about our organizations that we should have been examining all along. But we didn’t see, chose not to see, or remained disconnected from the challenges people were confronting.
The Permission Paradox is a perfect example, because the behavior at its center, concealment, has nothing to do with technology at all. People have always hidden things at work. What is new is the scale, the speed, and what the hiding reveals.
The Evidence Is Not Subtle
The largest study conducted on this question to date comes from KPMG and the University of Melbourne. Their report, Trust, Attitudes and Use of Artificial Intelligence: A Global Study 2025, surveyed 48,340 people across 47 countries between November 2024 and January 2025. The finding that should surprise every leader is that 57 percent of employees admitted to hiding their use of AI at work and presenting AI-generated content as their own. Nicole Gillespie, professor of management at the University of Melbourne and one of the study’s authors, described the pattern as a concerning level of inappropriate, complex, and non-transparent AI use, driven by pressure to meet targets, fear of being left behind, and organizational restrictions on the tools themselves.
Other studies put the number lower, and the variance itself is instructive. Ivanti’s 2025 workplace research, which surveyed more than 6,000 office employees, found that 32 percent keep their AI use secret from their employer. Slingshot’s Digital Work Trends report landed at 45 percent. The differences reflect different samples, different industries, and different ways of asking the question. But no credible study has found the number to be small. Whether the true figure in your organization is one in three or more than half, the conclusion is the same: a substantial portion of your workforce is doing meaningful work in ways you can’t see.
The reasons people give for hiding are where the psychology becomes visible. In the Ivanti research, 36 percent of those concealing their AI use said they enjoyed the secret advantage it gave them over peers. They have mastered or at least become better than most at the new skill or prompt generation. Thirty percent feared their job might be cut if their employer understood how much of their work could be automated. A real and growing fear across the majority of the workforce. Twenty-seven percent reported experiencing impostor syndrome, the sense that the work was no longer fully theirs. And if ever questioned, were they able to offer an explanation, a justification, or discuss in detail what they put forward as their own work. In the Slingshot data, 34 percent worried that colleagues would see AI use as cutting corners, and 27 percent feared being judged or misunderstood. We have entered complicated times, and times for which there is no real historical precedent.
Let’s place those answers next to what employers believe. In the same Slingshot research, 60 percent of employers said they believe their employees are being honest about AI use. Nearly half of employers attributed whatever secrecy exists to job security fears. The employees themselves told a different story. They were not primarily afraid of being replaced. They were afraid of being judged. Afraid of being seen as less capable, less authentic, less deserving of the work product they were submitting. The gap between what leaders think is happening and what is actually happening is not a data problem. It is a trust problem wearing a data problem’s clothes.
Why They Reached for the Tool in the First Place
Before we examine why people hide their AI use, we should be honest about why they adopted it at all, because the answer is not ambition or related to some novelty. The answer, for a very large share of the workforce, is cognitive depletion.
Microsoft and LinkedIn’s 2024 Work Trend Index, which surveyed 31,000 people across 31 countries, found that 68 percent of employees struggle with the pace and volume of their work and 46 percent report feeling burned out. The same research found that 75 percent of knowledge workers were already using AI, and 78 percent of those users were bringing their own tools to work rather than waiting for their organizations to provide them. The report’s own framing was unambiguous: employees, overwhelmed and under duress, are turning to AI for relief. Microsoft’s subsequent research on what it called the infinite workday found that one in three employees say the pace of work over the past five years has made it impossible to keep up. Read those findings together and the shadow adoption story changes the character of what is underway entirely. This is not a workforce seeking a competitive edge. Today’s workforce is treading water, and AI is the thing to help them manage their workload.
Decades of cognitive research explain what that treading water does to the mind. Gloria Mark’s long-running studies of attention at the University of California, Irvine, have documented how fragmented modern knowledge work has become and how costly each interruption is, with workers requiring extended recovery time to return to deep focus after every disruption. I often write about the individual’s need to be vulnerable. To take time to recover, to regain one’s cognitive capacity, or risk burnout from depletion.
Sophie Leroy’s 2009 research on attention residue showed that switching between tasks leaves part of our cognitive capacity stuck on the prior task, degrading performance on the next one. A workday built from meetings, messages, and context switches is a workday that systematically depletes the very cognitive resources that substantive work requires. When someone in that condition discovers a tool that drafts the routine email, summarizes the sixty-page document, or produces the first pass of the analysis, what they have found is not a shortcut. They have found a way to buy back cognitive capacity, and in many cases, to buy back their evenings and weekends. Much of what we are calling shadow AI adoption is more accurately described as a private work-life balance program, designed and implemented by individuals because the organization never offered one up. Regular readers will recognize the lineage here: this is The Busyness Trapmeeting its first genuine pressure valve, and people are turning the handle or flipping the switch quietly.
Objectivity requires acknowledging the complicating evidence, and it’s significant. The Upwork Research Institute surveyed 2,500 workers and executives across the United States, United Kingdom, Australia, and Canada in 2024 and found a startling inversion: while 96 percent of C-suite leaders expected AI to boost productivity, 77 percent of employees using AI said the tools had added to their workload. Workers reported spending more time reviewing AI-generated content, more time learning the tools, and being asked to do more work precisely because AI made them more capable. Kelly Monahan, the institute’s managing director, concluded that introducing new technology into outdated work models fails to unlock its value. At first glance, this seems to contradict the relief narrative. Look closer, and it sharpens it. The Upwork data describes what happens when AI arrives as a mandate, layered on top of unchanged expectations, with productivity gains harvested by the organization the moment they appear. The shadow adopter’s experience is different in kind: she chose the tool, she controls how it is used, and critically, she keeps the time it saves. The relief is real precisely because it’s hidden. The moment the organization finds out, the saved ninety minutes become ninety minutes of additional capacity where additional tasks and responsibilities can be assigned. Workers understand this intuitively, and it may be the least discussed reason for concealment of all: disclosure converts a personal recovery of time into an organizational claim on it.
Concealment Is Rational. That Is the Uncomfortable Part.
The instinct of most leadership teams, upon learning these numbers, is to treat hidden AI use as a compliance failure. Tighten the policy. Require disclosure. Add monitoring. This instinct misreads the situation completely because it assumes the concealment is the dysfunction. It isn’t. The concealment is a rational response to the dysfunction and the lack of organizational leadership direction, which is the organization’s refusal to decide.
Organizational psychology has understood this pattern for a long time. Patricia Hewlin’s research on what she termed facades of conformity, published in the Academy of Management Review in 2003, documented how employees suppress their authentic views and behaviors when they perceive that revealing them carries social or professional risk. People don’t hide because they are dishonest. They hide because the environment has taught them that disclosure is unsafe. Amy Edmondson’s foundational work on psychological safety, which I discuss probably too often, but which is so central to the human factor, beginning with her 1999 study in Administrative Science Quarterly. She demonstrated the same principle from the other direction: people share what they are doing, including their mistakes and their experiments, only when they believe the response will be learning rather than punishment.
Consider the position of an employee in an organization with no clear AI guidance, which describes most organizations today, not those you read about in the daily hype of the news cycle. She has discovered that a tool makes her measurably better at her job. Her options are these. She can ask permission, which invites a no, and a no would now require her to become worse at her job than she knows how to be. She can disclose her use after the fact, which risks judgment, reclassification of her performance, or in the worst case, the conclusion that her role can be automated. Or she can continue quietly, deliver excellent work, and say nothing. We can call the third option dishonest if we like. But we should be honest with ourselves about which option we would choose in her position. The organization built the maze. She is simply walking the only path through it that doesn’t end in a wall and one that protects her sanity (and cognitive resources).
This is what makes the Permission Paradox a true paradox. The absence of permission was supposed to be neutral, a pause while leadership figured things out. Instead, the absence of permission actively manufactures concealment. Every week the organization declines to decide, it teaches more employees that the safest course is silence. The organizations most worried about ungoverned AI use are, through their own indecision, producing the most of it.
The Inverse of Resistance
Regular readers know that one of the central arguments of this newsletter, and of The Truth About Transformation, is that resistance is a signal, not an obstacle. When people resist change, they are communicating something true about identity, competence, and loss, and leaders who listen to those signals navigate transformation far more successfully than leaders who try to overpower it.
The Permission Paradox presents us with resistance’s mirror image, and it’s just as rich a signal. These employees are not resisting change. They are changing faster than the organization permits. They have adopted, integrated, and normalized a new way of working entirely on their own, with no change management program, no training budget, no communications campaign, and no executive sponsor. Decades of transformation literature have been devoted to the question of how to get people to change. Almost none of it addresses what to do when people change ahead of you, in secret, and your formal transformation program is now trailing behind a shadow transformation that is already complete.
Think about what this means for the standard transformation playbook. We measure readiness with precision. We survey sentiment, assess capability, build adoption curves. And yet the KPMG data suggests that in many organizations, the most significant technology adoption of the decade is happening entirely outside those instruments. The dashboards show a workforce waiting to be transformed. The reality is a workforce that has already transformed itself and has concluded that telling you is not worth the risk. We have built elaborate systems for measuring whether people are ready to change, and almost nothing for detecting that they already have. Perhaps that ability will remain elusive in the pace of change we now find ourselves in.
The Cost of Not Knowing
It would be a mistake to read this issue as a celebration of shadow adoption, because the costs of the Permission Paradox are real for any and all organizations, and they compound.
The first cost is governance. The KPMG global study found that significant numbers of employees have entered sensitive organizational information into public AI tools, often without understanding where that data goes or how it is retained. When use is hidden, it is also unvetted. No security review, no legal assessment, no evaluation of whether the tool’s outputs meet the organization’s standards for accuracy. Every concealed workflow is a workflow no one has examined.
The second cost is organizational learning. When the senior director I described at the opening keeps her team’s methods private, the eleven people on her team get better, and the other four hundred people in her organization don’t. Hidden adoption traps the most valuable operational knowledge in the organization, the knowledge of what actually works, inside silos of secrecy. The organization is learning, but it’s learning the way a group of humans learns when every individual keeps their discoveries to themselves.
The third cost is the most corrosive, and it is the one leaders consistently underestimate. Every act of concealment is a small withdrawal from the trust account between employees and leadership. The employee who hides her methods today finds it easier to hide other things tomorrow. The manager who suspects his team is using unapproved tools but prefers not to know is practicing a willful blindness that his team can feel. Over time, the Permission Paradox doesn’t just hide AI use. It normalizes the broader premise that what leadership knows about how work happens and how work actually happens are two different things, and that this is fine. No organization can transform on top of that premise. The foundation from which any change or transformation initiatives come is not based on facts or reality.
What Leaders Owe Their People Here
The Permission Paradox isn’t a process failure that a better process will fix. The default thought process is that an organization can simply fix the failure with a structure, a policy, or a new process. The reason is that the failure results from a trust condition, and trust conditions respond to posture, not procedure.
The posture begins with a decision, because the root of the paradox is organizational indecision. Leaders owe their people an actual answer to the question of what is permitted, even an imperfect and provisional one. A clear and evolving policy that says here is what you may use, here is what you may not, here is the information you can submit, and here is the information you cannot, and here is how we will revisit this as we learn, removes the ambiguity that makes concealment the rational choice. The organizations handling this well aren’t the ones with the most restrictive policies or the most permissive ones. They are the ones whose people no longer have to guess. Because when there is a vacuum of leadership direction or some precedent to fall back on, individuals will fill the vacuum with their own choice.
The posture continues with what we might call disclosure amnesty. If 32 to 57 percent of your workforce is already using these tools in secret, then any honest path forward begins by making it safe to say so. That means explicitly separating disclosure from punishment. It means treating every surfaced use case as diagnostic information about unmet needs, inadequate tools, or workflow friction that your formal systems failed to detect. It may also expose where your leadership is missing from the conversation.
The employee who found a way to cut a six-hour task to ninety minutes has handed you operational intelligence that no consultant could have produced. Whether she ever hands it to you depends entirely on what she believes will happen to her when she does.
And the posture rests, finally, on a harder piece of self-examination. If your people are hiding their best work and the way they produce it from you, the question worth sitting with isn’t what is wrong with them. It’s what they have learned about you. Concealment at this scale is not a character flaw distributed across half the workforce. It is a verdict on the environment, rendered independently by thousands of reasonable people who all reached the same conclusion about what was safe to reveal. Leaders who can absorb that verdict without defensiveness, and treat it as the trust diagnostic it is, will find that the Permission Paradox dissolves faster than any policy could dissolve it.
The Question Underneath
Last month, in The Readiness Illusion, we examined the gap between what AI systems can do in demonstrations and what organizations are actually ready to deploy, and I argued that capability and readiness are not the same thing. The Permission Paradox is what happens when that same gap goes unresolved at the individual level. The organization can’t decide what it’s ready for, so every employee decides privately, and the sum of those private decisions becomes the organization’s real AI strategy, unwritten, unexamined, and invisible to the people nominally in charge of it.
So the question for leaders is not whether your people are using AI. The evidence says they are, at scale, right now, whether or not you have approved a single tool. The question is whether your organization has earned the kind of trust that lets its people tell you. That answer will not appear on any dashboard. But it is, right now, the most important readiness metric you have.
Related Reading
The Permission Paradox: Why Your People Are Hiding the Very Things You Say You Want (Issue 269)
Directly Referenced in This Issue
The Readiness Illusion: Why the AI Agent Era’s Loudest Claims Outrun the Evidence
The preceding issue, which argues that capability and readiness are not the same thing. The Permission Paradox extends this argument to the individual level, where organizational indecision forces employees to resolve the readiness question privately.
The Busyness Trap: Why We Wear Exhaustion as a Status Symbol (Issue 250)
The cognitive depletion narrative in The Permission Paradox traces directly to this earlier issue. Shadow AI adoption is, in many cases, individuals finding their own pressure valve for the exhaustion cycle described here.
The Truth About Transformation: Revised and Expanded
The Permission Paradox presents the mirror image of the signal: what happens when people change faster than the organization permits.
Ideas and Innovations Newsletter Issues
The Algorithmic Mirror: What AI Reveals About How We Actually Think and Decide
Explores how AI functions as a diagnostic lens for organizational behavior, the same framing that The Permission Paradox builds on when it argues that AI is not the subject but the lens.
Decision Theater: When People and Organizations Mistake Motion for Commitment
Organizational indecision is the root of the Permission Paradox. Decision Theater examines the broader pattern of organizations performing the rituals of decision-making without actually exercising authority.
The Mental Overload of Modern Leadership: Why Today’s Executives Are Burning Out Differently
The cognitive depletion research cited in The Permission Paradox connects directly to this earlier examination of how leadership demands have outpaced human cognitive architecture.
When Change Champions Burn Out: The Hidden Cost of Driving Change and Transformation
Explores the cognitive burden that falls on individuals who carry transformation forward, the same population most likely to become shadow AI adopters seeking relief.
The Loyalty Trap: When Commitment Becomes a Cage
Examines how organizational loyalty can become a mechanism that suppresses authentic behavior, a dynamic closely related to the facades of conformity and concealment patterns described in The Permission Paradox.
The Meeting after the Meeting: Why Your Strategies and Plans Keep Losing Support
Explores the gap between what is said in formal settings and what people actually believe, a trust and disclosure dynamic that runs parallel to the concealment patterns in The Permission Paradox.
Artificial Understanding: The Intelligence We Built and the Comprehension We Didn’t
Examines the gap between AI capability and genuine organizational comprehension, providing context for why employees adopt AI tools that organizations have not yet learned to govern.
Human Factor Podcast Episodes
Season 2, Episode 026: The Readiness Illusion
The companion podcast episode to the newsletter issue directly referenced in The Permission Paradox. Explores the gap between AI capability and organizational readiness in greater depth.
Season 2, Episode 019: Structural Silence: Why Organizations Train People Not to Speak
Directly relevant to the concealment dynamics at the heart of The Permission Paradox. Examines how organizations systematically teach employees when not to speak up and the burnout that results from sustained silence.
Explores AI as a diagnostic lens for organizational behavior, the same framing that informs The Permission Paradox’s argument that concealment reveals more about the organization than it does about the technology.
Season 2, Episode 027: Transformation Theater: Why Organizations Perform Change Instead of Making It
Connected to the Decision Theater concept and the organizational indecision that produces the Permission Paradox. Examines performance without commitment in transformation contexts.
The impostor syndrome and identity threat dimensions of The Permission Paradox connect directly to this episode’s examination of how professional identity becomes disrupted when expertise is challenged by new tools.
Season 1, Episode 009: Transformation Fatigue: When Your Organization Can’t Absorb More Change
The cognitive depletion argument in The Permission Paradox builds on the transformation fatigue research explored in this earlier episode, where exhaustion becomes the precondition for shadow adoption.
Measuring What Matters Series
The Permission Paradox identifies trust as the core issue beneath concealment. This Measuring What Matters article provides the framework for actually measuring the trust conditions that produce or dissolve the paradox.
The Identity Problem in Measurement
Examines how professional identity shapes what organizations choose to measure and what they exclude. The Permission Paradox describes a workforce that has transformed itself outside the measurement systems, making the identity lens on measurement directly relevant.
Why Transformation Dashboards Lie
The Permission Paradox argues that dashboards no longer describe how work actually gets done. This series’ article examines the structural reasons why transformation metrics misrepresent reality.
The Organizational Memory Problem
When shadow adoption traps operational knowledge inside silos of secrecy, the organization loses the ability to learn from its own experience, the core problem examined in this series’ article.
Transformation Psychology
Why AI Adoption Resistance in the Workplace Is a Leadership Problem, Not a Technology Problem
The foundational article argues that AI resistance is rooted in identity, competence, and trust rather than technology literacy. The Permission Paradox extends this by examining what happens when resistance inverts, and people adopt ahead of the organization.
The Partnership Posture: A Working Frame for Humans and AI
Offers a constructive framework for the human and AI relationship that the Permission Paradox argues most organizations have failed to articulate, leaving employees to define the relationship on their own.
Sources
KPMG and the University of Melbourne, Trust, Attitudes and Use of Artificial Intelligence: A Global Study 2025.
Microsoft and LinkedIn, 2024 Work Trend Index Annual Report, AI at Work Is Here. Now Comes the Hard Part.
Microsoft WorkLab, Breaking Down the Infinite Workday, 2025.
The Upwork Research Institute, AI-Enhanced Work Models, 2024.
Ivanti, 2025 Workplace Technology Research.
Slingshot, Digital Work Trends Report.
Gloria Mark, University of California, Irvine and Sophie Leroy, Why Is It So Hard to Do My Work? The Challenge of Attention Residue When Switching Between Work Tasks, Organizational Behavior and Human Decision Processes, 2009.
Patricia Hewlin, And the Award for Best Actor Goes to…Facades of Conformity in Organizational Settings, Academy of Management Review, 2003.
Amy Edmondson, Psychological Safety and Learning Behavior in Work Teams, Administrative Science Quarterly, 1999.
This is the part leaders often miss: hidden AI use is not only a technology problem. It is trust data.
When people hide how the work is actually getting done, the better question is not “why are employees being difficult?”
It is: “what have they learned is unsafe, pointless, or punished to say out loud?”
Policy matters. But culture decides whether the truth reaches the room.