Deep Dive: The Hidden World of Shadow AI

Olivia Carter

Good afternoon, Mark. Today, we’re pulling back the curtain on something that’s quietly reshaping the workplace. It’s not about a new tool or platform—it’s about the fact that many people are already using generative AI at work, but almost no one is being honest about it. There’s this gap between how work is getting done and what anyone is actually acknowledging out loud.

Mark Putnam

Yeah, and what’s wild is, this isn’t just some theory. It’s backed up by data. That KPMG–University of Melbourne study from earlier this year surveyed nearly 50,000 workers across 47 countries. Fifty-seven percent admitted they’re hiding their AI use. Not disclosing, not clarifying—just quietly using it. And nearly half are breaking their own company’s rules while doing it.

Olivia Carter

That’s right. This isn’t a handful of bad actors. This is a cultural shift that’s happening under the radar. Tools like ChatGPT, Claude, and Gemini aren’t just novelty apps anymore—they’ve become part of how knowledge work gets done. People are drafting emails, building reports, summarizing documents, even writing code with AI. But they’re doing it quietly, in personal tabs or side windows, and not bringing it into the open.

Mark Putnam

And that’s what makes this so important. GenAI is already embedded in daily workflows, but in most organizations, it’s invisible. Managers don’t know when it’s being used. Peers don’t see the process behind polished deliverables. The productivity shows up, but the method disappears.

Olivia Carter

And that invisibility is not neutral. It creates a trust gap. When the process behind the work is hidden, the assumptions about effort, originality, and intent all start to erode. The issue isn’t the technology itself—it’s the silence. The silence is what breaks transparency. It’s what makes collaboration fragile. And it’s what prevents leaders from actually understanding what’s happening inside their teams.

Mark Putnam

That’s the real headline here. The story of AI at work in 2025 isn’t just about rapid adoption. It’s about concealment. People are using these tools more and more, but they’re doing it in isolation, without discussion, policy guidance, or shared norms. That’s what makes this such a tricky leadership challenge. It’s not about blocking AI. It’s about breaking the silence.

Olivia Carter

So let’s talk about why this really matters—because it’s not just about how people are using AI. It’s about whether they’re being honest about it. That shift, from usage to honesty, is massive. We’re not in a stage where we’re figuring out if AI can help. We’re in a stage where we’re figuring out whether anyone feels safe admitting they’re using it.

Mark Putnam

Right, and the numbers paint a pretty clear picture. According to that same KPMG study, 66 percent of workers don’t even verify the output they get from AI tools. That’s two out of three people relying on machine-generated content without checking it. And about a third of workers now say that shadow AI use is normal on their teams. Not occasional. Not experimental. Normal.

Olivia Carter

And that creates real cultural fractures. You’ve got some employees quietly automating parts of their job—maybe making their work look faster, sharper, more consistent—while others are following the traditional rules, doing it the long way, and possibly falling behind. That’s not just a productivity imbalance. It starts to erode the whole idea of fairness.

Mark Putnam

That’s right. When people don’t know how work is getting done, trust breaks down. If I don’t know whether your results came from skill, AI, or a shortcut, how do I know what to expect next time? And how do I calibrate my own work when the playing field is basically invisible?

Olivia Carter

It’s like we’ve introduced a hidden variable into every project and every team interaction. And it’s not the AI itself that’s the issue—it’s the lack of transparency around it. When processes are hidden, expectations become unclear. When expectations are unclear, collaboration starts to suffer.

Mark Putnam

And that’s really the deeper risk here. If everyone is keeping quiet about how their work is getting done, it creates a culture of silence. People stop sharing. They stop asking for help. They stop trusting that what they’re seeing is what they’re getting.

Olivia Carter

It becomes harder to build real teams. Because collaboration is built on shared understanding. And when GenAI is being used behind closed doors, that understanding slowly disappears.

Mark Putnam

Let’s talk about the part that’s easy to misunderstand—why people are hiding their AI use in the first place. It’s tempting to chalk it up to laziness or cutting corners, but that really misses the point. The dishonesty isn’t moral, it’s structural. People are responding to mixed signals. The rules are unclear, there’s very little training, and the pressure to keep up is intense. So, if AI helps you meet the deadline, you use it—but quietly.

Olivia Carter

That’s right. And what makes it worse is that AI use is often implicitly encouraged. Teams are expected to be faster, more efficient, more responsive. But if you’re open about how you did it—if you say “I used ChatGPT to help draft this”—you might get side-eyed or have your work taken less seriously. There’s this weird stigma around admitting you got help, even if the results are good.

Mark Putnam

And that’s not just anecdotal. The University of Arizona ran a whole series of experiments on this. They found that when people disclosed their AI use, trust actually went down. Students trusted professors 16 percent less when they learned AI graded their work. Investors trusted companies less when they found out an ad was made with AI. Clients trusted designers less when they admitted to using tools like Midjourney.

Olivia Carter

It’s wild. Even when the outcomes were solid, just knowing that AI was involved made people uneasy. And that creates what researchers call a “transparency penalty.” You’re honest, and you get dinged for it.

Mark Putnam

And here’s the kicker—if you’re not honest and someone finds out, the trust drop is even worse. So now you’re stuck in this brutal double bind: say nothing and risk a blow-up later, or be transparent and get quietly dismissed now.

Olivia Carter

So what do most people do? They stay silent. It’s the safest play in a broken system. Not because they’re trying to deceive anyone, but because the system itself is sending mixed signals. Use AI, but don’t talk about it. Be efficient, but look human. Innovate, but stay within unwritten boundaries.

Mark Putnam

And that’s how you end up with this culture of concealment. Everyone is navigating this gray zone, making decisions in isolation, unsure of what’s acceptable. Over time, silence becomes the default. Not because it’s strategic—but because it feels like survival.

Olivia Carter

Right.

Mark Putnam

Let’s zoom out for a second. This isn’t just a quirky trend or a passing workplace habit—it’s a serious governance failure. When employees are using AI tools like ChatGPT or Copilot without oversight or disclosure, leaders lose visibility into how work is actually getting done. And if you don’t know how the work happens, you can’t manage the risks, you can’t enforce standards, and you certainly can’t evaluate performance on a fair playing field.

Olivia Carter

That’s right. And that loss of visibility creates some very real consequences. For starters, compliance starts to break down. Employees are pasting sensitive or confidential data into public AI tools without approval. That’s not just a policy violation—it’s a legal and security risk. And then there’s the inequity issue. Some employees are quietly automating major parts of their workflow and getting praised for how fast they move, while others—who might be doing things by the book—fall behind. It’s not a fair system, and it’s not sustainable.

Mark Putnam

And the quality control issues are no small thing either. A lot of people don’t double-check what AI tools are producing. That means factual errors, misinterpretations, or even hallucinated content can make it into deliverables unchecked. By the time someone notices, it’s often too late. The damage is done.

Olivia Carter

And we’re also starting to see the effect on collaboration. One in five workers says that AI is making their teams less collaborative. That’s a huge red flag. When people stop asking each other for help or stop walking through problems together because they’re relying on private AI chats instead, you lose the learning, the shared understanding, the team cohesion. The tools might be saving time, but they’re also cutting human connection out of the process.

Mark Putnam

Yeah, and that’s where the metaphor really hits. It’s like there’s this hidden layer of work happening just below the surface. It’s fast. It’s efficient. But it’s not transparent, and it’s not accountable. Leaders are essentially flying blind while this parallel process reshapes how work gets done. And if they don’t address that soon, it’s going to come back to bite them—hard.

Olivia Carter

So let’s get practical—what should leaders actually be doing about this? Because the answer isn’t more technology. It’s not about buying the next AI dashboard or locking down employee laptops. The real challenge here is cultural. This is a leadership problem, not a tooling problem. If people are hiding how they work, it’s because the culture you’ve built makes honesty feel risky.

Mark Putnam

That’s right. The first step is to write clear, thoughtful AI policies that focus on responsible use, not just restrictions. When you don’t give people guidance, they make their own rules—or worse, they stop talking altogether. A total ban just drives everything underground. It doesn’t change behavior, it just hides it better.

Olivia Carter

And part of that guidance means giving people vetted tools. If you want employees to stay within the lines, you’ve got to draw those lines in the first place. That means providing access to approved GenAI platforms that are secure, trustworthy, and integrated into existing workflows. Don’t make them choose between getting their job done and following the rules.

Mark Putnam

And then there’s the training gap. Most employees—even the ones using AI every day—have had no real training. They’re just figuring it out on their own, pulling prompts off Reddit, or relying on word of mouth. That’s a recipe for both risk and burnout. Training shouldn’t just be for IT or early adopters. It should be built into onboarding, professional development, even performance reviews.

Olivia Carter

But maybe the most important thing leaders need to create is psychological safety. People have to feel like they can say, “I used ChatGPT to get started on this,” or “I had Copilot outline the first draft,” without fearing judgment or penalty. That openness doesn’t happen by accident. Leaders have to model it. They need to say out loud when they use these tools themselves. Not just tolerate AI use from their teams, but normalize it.

Mark Putnam

Because here’s the thing—if you want transparency, you have to create the conditions for it. That means clarity, trust, safety, and accountability. Otherwise, you’re just hoping your employees will tell you the truth while giving them every reason not to.

Olivia Carter

So let’s pull all of this together. We’re not just in a phase of GenAI adoption—we’re in a phase of GenAI concealment. These tools are already part of how work gets done. But instead of talking about it, most employees are staying quiet, navigating gray areas, and making decisions in the dark.

Mark Putnam

And that silence? It’s not harmless. It’s shaping culture in ways we don’t always see right away. Leaders need to ask a different question—not just “Who’s using AI?” but “Why do so few people feel safe saying that they are?” Because that discomfort, that hesitation, that fear of judgment—that’s the cultural signal you need to be listening to.

Olivia Carter

That’s right. The future of AI in the workplace isn’t going to be decided by which tools we buy or which vendors we choose. It’s going to be shaped by the culture we create around those tools. A culture where transparency is normal. A culture where innovation isn’t punished with suspicion. A culture where leaders lead with clarity, not silence.

Mark Putnam

So here’s our challenge to you: confront the silence before it defines you. Start the conversations your team isn’t having. Create the space where people can be honest about how they work. Because if you don’t shape the culture around AI, it will shape itself—and you probably won’t like the results.

Olivia Carter

Mark, thanks for unpacking this with me. It’s a complex topic, but one I think every leader needs to sit with—especially right now.

Mark Putnam

Always a pleasure, Olivia. And thanks to all of you for listening. We’ll be back soon with more insights at the intersection of leadership, technology, and workplace culture.

Olivia Carter

See you next time. And until then, don’t just adopt the tools—build the culture.