There’s a double bind facing employees when it comes to artificial intelligence. They’re expected to use it to be more productive, yet judged on how they do so. Recent Deloitte research found that nearly two-thirds (63 percent) of the UK workforce has now used generative AI, a third of those users do so without their employer’s knowledge, and one in ten have used an AI tool that their employer has banned or would disapprove of. Separate Deloitte figures found show that one in six are even paying for AI tools they use at work out of their own pocket.
The instinctive response for some employers is to treat this as a governance problem, with tighter policies, blocked sites and more monitoring. The National Cyber Security Centre (NCSC) has also warned, with good reason, about the risks of shadow AI going unchecked.
Yet framing this purely as rule-breaking misreads what’s happening. It doesn’t address why employees feel they need to hide AI use in the first place, and it risks making things worse.
The mental toll of concealment
A common assumption is that employees hide their AI use because they’re cutting corners or passing off AI-assisted work as their own. While there is some truth in that, it’s far from the whole picture.
Much of the urge to conceal is rooted in anxiety. There’s the fear of being replaced, with two-thirds of Deloitte’s weekly AI users worried their managers will think AI can do their job. There’s the fear of being judged, with nearly a quarter citing the stigma attached to using it. And there’s the sense of being left to work it out alone – only around a third say their leaders talk about AI with a good understanding, and half have had no training.
These fears aren’t unfounded. In one US experiment, Atlassian found that people rated a colleague who disclosed their AI use as ten times lazier than one who didn’t, and were far less likely to recommend them for a high-visibility project – even when the work was identical.
The anxiety doesn’t ease as people get more out of AI, either. Anthropic’s research found that people in the roles most exposed to AI are three times more likely to worry about being displaced by it than those in the least exposed roles, and those reporting the biggest productivity gains from AI are also among the most concerned.
It also falls hardest on younger workers, who are the most worried about displacement. These are the people who most need guidance, because they’re still building the skills and judgement their careers will depend on. If they don’t feel safe talking about how they use AI, they’re less likely to get that guidance, and less likely to learn when to rely on AI and when not to.
The price of secrecy
Heavy-handed bans and surveillance tend to feed these fears rather than resolve them, which only pushes AI use further out of sight. As one analysis of shadow AI put it: “Bans rarely eliminate usage. They simply drive it underground.”
Once AI use goes underground, organisations lose the openness they most need during a period of rapid change. Knowledge stops spreading, because nobody shares what works. Mistakes go unchallenged, because nobody admits how a piece of work was produced. And the people designing roles, training, systems and workspaces lose sight of how work is actually being done.
Even the NCSC isn’t telling people to block every tool. It advises organisations to adopt “a positive cyber security culture” with an open dialogue. Deloitte reached a similar conclusion, arguing that the answer isn’t simply stricter rules. When a national security body and a global firm both say openness beats control, HR leaders have a clear mandate to act.
What employers can do
Many people use AI freely in their personal lives, yet hold back or hide it at work. This gap tells us that the issue isn’t the technology itself, but the way it’s embedded into the workplace. Around 35 conditions need to be met in order for workplace AI adoption to be successful, and half of those are cultural – so clearly this is not simply a technological problem. If we can fix the culture, we can create the foundations for good-quality, open AI use.
That starts at the top. Nothing reduces stigma faster than senior people talking openly about how they use AI, including where it helped and where it got things wrong. If leaders keep their own use hidden, employees will draw the obvious conclusion.
It also means looking beyond ‘productivity’ and treating AI as a human development issue. Employees need help understanding how their roles are changing, where their judgment matters and which human qualities are worth protecting. Used well, AI clears away repetitive tasks to create more room for creativity and big-picture thinking. Used badly, it replaces the very practices through which people build their human experience and expertise.
None of this means employers have to give up control. Those that want to manage artificial intelligence more closely can, but rules only work if people first feel psychologically safe, engaged and looked after. Without those conditions, tighter control will simply push AI use further underground. It’s up to us as employers to put the clarity, dialogue and support in place to ensure that no one in our organisation feels they have to choose between being productive and being trusted.
Gethin Nadin is chief advisory officer at Benifex and author of Heart in the Age of the Mechanical. Gethin is an award-winning psychologist, bestselling HR author, and in 2026 was named one of the UK’s Top 100 Most Influential People. Gethin has been awarded more than ten awards since 2024 for his work in influencing the global HR industry, most recently awarded double Gold at the prestigious Stevie’s awards in New York in recognition of a career dedicated to human resources. He was also awarded a lifetime achievement award by the UK Employee Experience Awards in 2025 and named PwC’s ‘Outstanding Contribution to the Industry 2024’. Gethin is a Fellow at the historic King’s College London, King’s Business School and Chair of the Westminster think tank, The Policy Liaison Group on Workplace Wellbeing.
October 7, 2026
Employees aren’t hiding their AI use to cut corners. They’re hiding it because they’re afraid
by Gethin Nadin • AI, Comment
The instinctive response for some employers is to treat this as a governance problem, with tighter policies, blocked sites and more monitoring. The National Cyber Security Centre (NCSC) has also warned, with good reason, about the risks of shadow AI going unchecked.
Yet framing this purely as rule-breaking misreads what’s happening. It doesn’t address why employees feel they need to hide AI use in the first place, and it risks making things worse.
The mental toll of concealment
A common assumption is that employees hide their AI use because they’re cutting corners or passing off AI-assisted work as their own. While there is some truth in that, it’s far from the whole picture.
Much of the urge to conceal is rooted in anxiety. There’s the fear of being replaced, with two-thirds of Deloitte’s weekly AI users worried their managers will think AI can do their job. There’s the fear of being judged, with nearly a quarter citing the stigma attached to using it. And there’s the sense of being left to work it out alone – only around a third say their leaders talk about AI with a good understanding, and half have had no training.
These fears aren’t unfounded. In one US experiment, Atlassian found that people rated a colleague who disclosed their AI use as ten times lazier than one who didn’t, and were far less likely to recommend them for a high-visibility project – even when the work was identical.
The anxiety doesn’t ease as people get more out of AI, either. Anthropic’s research found that people in the roles most exposed to AI are three times more likely to worry about being displaced by it than those in the least exposed roles, and those reporting the biggest productivity gains from AI are also among the most concerned.
It also falls hardest on younger workers, who are the most worried about displacement. These are the people who most need guidance, because they’re still building the skills and judgement their careers will depend on. If they don’t feel safe talking about how they use AI, they’re less likely to get that guidance, and less likely to learn when to rely on AI and when not to.
The price of secrecy
Heavy-handed bans and surveillance tend to feed these fears rather than resolve them, which only pushes AI use further out of sight. As one analysis of shadow AI put it: “Bans rarely eliminate usage. They simply drive it underground.”
Once AI use goes underground, organisations lose the openness they most need during a period of rapid change. Knowledge stops spreading, because nobody shares what works. Mistakes go unchallenged, because nobody admits how a piece of work was produced. And the people designing roles, training, systems and workspaces lose sight of how work is actually being done.
Even the NCSC isn’t telling people to block every tool. It advises organisations to adopt “a positive cyber security culture” with an open dialogue. Deloitte reached a similar conclusion, arguing that the answer isn’t simply stricter rules. When a national security body and a global firm both say openness beats control, HR leaders have a clear mandate to act.
What employers can do
Many people use AI freely in their personal lives, yet hold back or hide it at work. This gap tells us that the issue isn’t the technology itself, but the way it’s embedded into the workplace. Around 35 conditions need to be met in order for workplace AI adoption to be successful, and half of those are cultural – so clearly this is not simply a technological problem. If we can fix the culture, we can create the foundations for good-quality, open AI use.
That starts at the top. Nothing reduces stigma faster than senior people talking openly about how they use AI, including where it helped and where it got things wrong. If leaders keep their own use hidden, employees will draw the obvious conclusion.
It also means looking beyond ‘productivity’ and treating AI as a human development issue. Employees need help understanding how their roles are changing, where their judgment matters and which human qualities are worth protecting. Used well, AI clears away repetitive tasks to create more room for creativity and big-picture thinking. Used badly, it replaces the very practices through which people build their human experience and expertise.
None of this means employers have to give up control. Those that want to manage artificial intelligence more closely can, but rules only work if people first feel psychologically safe, engaged and looked after. Without those conditions, tighter control will simply push AI use further underground. It’s up to us as employers to put the clarity, dialogue and support in place to ensure that no one in our organisation feels they have to choose between being productive and being trusted.
Gethin Nadin is chief advisory officer at Benifex and author of Heart in the Age of the Mechanical. Gethin is an award-winning psychologist, bestselling HR author, and in 2026 was named one of the UK’s Top 100 Most Influential People. Gethin has been awarded more than ten awards since 2024 for his work in influencing the global HR industry, most recently awarded double Gold at the prestigious Stevie’s awards in New York in recognition of a career dedicated to human resources. He was also awarded a lifetime achievement award by the UK Employee Experience Awards in 2025 and named PwC’s ‘Outstanding Contribution to the Industry 2024’. Gethin is a Fellow at the historic King’s College London, King’s Business School and Chair of the Westminster think tank, The Policy Liaison Group on Workplace Wellbeing.