+44 (0) 203 633 1790

Article

Is your team hiding its AI use? What UK surveys show

Deloitte UK's survey of 25,000 workers, published in September 2026, found 31% of generative AI users use it without their employer's knowledge. PwC UK found 85% of senior executives and 73% of managers had used AI at work, against 35% of non-managers. The divide in these surveys follows seniority, and the hiding points to a trust gap that managers can close by making AI use easy to discuss.

Nearly a third of UK workers who use generative AI do so without their employer's knowledge: 31%, according to Deloitte UK's GenAI Workforce Survey, published on 16 September 2026. The gap that matters most in the UK surveys is between senior and junior staff, and a team that hides its AI use has a trust problem before it has a technology problem.

The figures in this article were the latest available in mid November 2026. Deloitte says it will repeat its survey about every six months, so newer figures should follow.

31%of generative AI users say they use it without their employer's knowledgeDeloitte UK, 16 September 2026
85%of UK senior executives had used AI at work, against 35% of non-managersPwC UK, 1 October 2026
64%of weekly generative AI users worry managers will think AI can do their jobsDeloitte UK, 16 September 2026
59%of Gen Z had used AI platforms such as ChatGPT, against 13% of Baby BoomersCommunication Habits and Preferences 2024

Who uses AI at work: the young or the senior?

In the PwC data, seniority matters more than anything else the release reports. PwC UK's Hopes and Fears survey, published on 1 October 2026, included 2,023 UK workers. It found that 56% had used AI at work in the past year and 19% used it daily. Use rose sharply with rank: 85% of senior executives and 73% of managers had used it, against 35% of non-managers.

Used AI at work in the past year, by seniorityShare of UK workers in each group
  • Senior executives85%
  • Managers73%
  • Non-managers35%

Base: 2,023 UK workers (the release does not give a base for each group) Source: PwC UK Hopes and Fears survey, published on 1 October 2026

Neither PwC nor Deloitte gives an age or generation breakdown in its release, so these surveys cannot say whether age matters once rank is taken into account. What they do show is that rank matters. The popular picture of junior staff racing ahead of cautious managers does not match the seniority pattern PwC reports, in which the highest use sits at the top of the organisation.

How many people are hiding their AI use?

Deloitte's survey covered 25,000 UK workers aged 18 to 70, with fieldwork by Ipsos UK from 7 May to 10 June 2026. Of the 63% who said they use generative AI for work (16,749 respondents), 31% said they do so without their employer's knowledge. Deloitte calls this shadow AI. The 31% is a share of those users, not of all workers.

Other figures in the same release point to why people stay quiet. Around half of employees who use generative AI at work said they had received no formal training. Deloitte reported that 65% of users saw a lack of convincing leadership on how it should be used. Nearly a quarter of all respondents (23%) thought there was a stigma attached to using it at work, and 64% of weekly users worried their managers would think AI could do their jobs. Deloitte's Hayley McKelvey said that workers who feel a stigma are more likely to conceal their use.

PwC's release adds a smaller signal. It found 16% of UK workers fearing negative consequences if they experiment with AI, and 13% confused about their company's policy. Both surveys come from professional services firms that advise on AI, which is worth bearing in mind, but the findings point the same way.

Does our own survey show an age gap in AI use?

It does, but it measures something different. In our 2024 survey, Communication Habits and Preferences 2024, 59% of Gen Z had used AI platforms such as ChatGPT, against 48% of Millennials and 13% of Baby Boomers. The survey had 4,067 respondents and fieldwork ran from July to September 2024.

Had used AI platforms such as ChatGPT, by generationShare of each generation, 2024
  • Gen Z59%
  • Millennials48%
  • Baby Boomers13%

Base: 4,067 people in the UK Source: Communication Habits and Preferences 2024

These figures cannot be set beside PwC's. Ours asks whether people had ever used AI platforms, covers everyone aged 8 and over rather than only workers, and was collected two years earlier. Our Baby Boomer group, aged 60 to 78 at the time, includes people no longer in work. PwC asks whether employed people used AI at work in the past year, and splits them by rank. Taken together, they suggest age is only part of the picture at work: in PwC's figures, seniority makes a large difference to who uses AI.

That is the pattern we describe in Patterns, not pigeonholes: birth year gives a hint, but job, access and training often explain more. Our piece on AI and Gen Z's first jobs covers the graduate side of the story, so this article stays with teams already in post.

Why is hidden AI use a communication problem?

When a person keeps a tool secret, they are telling you something about how safe it feels to be open. A junior colleague may fear being seen as someone who cuts corners or whose job can be automated. A senior colleague may not want to admit they are still learning. In both cases the manager does not get to hear what the tool is good at, where it goes wrong, or whether it is being used with client data it should never touch.

The risk for employers is practical as well as relational. Deloitte found that nearly half (46%) said they used free tools at work, with no base given in the release. Without a conversation, nobody agrees what can be pasted into them.

How can managers make AI use easier to talk about?

Start by saying it is allowed to be discussed. A short, plain statement from a manager works better than a policy document: you can tell me when you use AI, and nobody will be marked down for it. Then back it up. If the first person who admits to using it gets a lecture, the rest of the team will stay quiet.

Agree what is acceptable in writing, in language people can follow. Cover which tools are approved, what information must never go into them, when to tell a client or colleague that AI helped, and who checks the output. Ask people what they already do before you set the rules, because the rules will otherwise miss real practice.

Go first. PwC's figures suggest senior people are among the heaviest users, so show your own use, including something that went wrong. It tells the team that mistakes with AI are normal and can be discussed.

Offer training to everyone, not only the people you assume need it. With around half of users in Deloitte's survey untrained, assumptions about who is skilled are likely to be wrong in both directions.

Use two-way learning. A junior colleague may know a tool well while a senior one knows what good work looks like and where AI output needs checking. Our guide to reverse mentoring sets out how to pair them so that each learns from the other. Our podcast episode From hype to workflow: making AI actually useful takes the practical view.

Frequently asked questions

How many UK workers use AI without telling their employer?

Deloitte UK's GenAI Workforce Survey, published on 16 September 2026, found that 31% of generative AI users use it without their employer's knowledge. The survey covered 25,000 workers aged 18 to 70 and was fielded by Ipsos UK from 7 May to 10 June 2026. The 31% is a share of users, who made up 63% of respondents.

Do younger workers use AI more than older workers?

It depends on what is measured. In our 2024 survey, 59% of Gen Z had used AI platforms, against 13% of Baby Boomers. In PwC UK's 2026 survey, use at work in the past year was 85% among senior executives and 35% among non-managers. PwC gives no age breakdown, and the two surveys measure different things, so they cannot be compared directly.

Why do employees hide their AI use?

Deloitte's release points to stigma, worry about job security and a lack of clear guidance. It reported that 64% of weekly users worry managers will think AI can do their jobs and that 65% of users saw a lack of convincing leadership on AI. PwC found 16% of UK workers fear negative consequences if they experiment with AI.

What should an AI policy for a team include?

Name the approved tools, say what information must never be entered, set out when to disclose AI help and say who checks the results. Write it with the people who already use AI, and review it as tools change.

Hear how leaders are bridging the generations on AI and everything else on the Generationally Speaking podcast.

Listen to the podcast

Listen: related podcast episodes

Invite Alastair to speak at your event

Contact
Get in touch