A new report from Imperial Lifelong Learning, released on 11 November 2026 and covered again on 12 November, puts numbers to something most employers already suspect. In a Censuswide survey of 2,000 UK employees across five high-growth sectors, nearly three in ten Gen Z workers (29%) said they use AI without their employer's knowledge. So did 18% of Millennials, 11% of Gen X and 6% of Baby Boomers.
It is tempting to treat that as a discipline problem: ban the unofficial tools, tighten the rules, and tell people to stop. Our view is simpler. Shadow AI in the workplace usually means people are trying to get work done without a sanctioned way to do it, or they are staying away from AI because they do not trust it. Both are skills and policy problems, and both are fixable.
What the report actually found
The report, Everyone but me: What UK employees told us about AI and skills, is based on a Censuswide survey of 2,000 UK employees, including business owners and excluding sole traders, carried out between 4 and 15 September 2026. Fieldwork set a minimum of 100 responses in each of five sectors: financial services; engineering, manufacturing and infrastructure; technology and digital; energy and sustainability; and healthcare and life sciences. Those findings describe workers in those sectors, not the entire UK workforce. Imperial also ran ten qualitative interviews with senior leaders between late September and mid-October 2026.
The confidence gap is the other headline. Almost all workers in the survey (97%) were confident they have the skills their current role will need in the next 12 months. At the same time, 59% saw a significant skills gap in their industry, and 75% of employers said the same. Regular users of AI were nearly twice as likely to recognise that gap.
The picture is uneven in other ways too. Imperial's 12 November news article notes uneven patterns of AI adoption across sectors, generations, levels of seniority and regions. One in five respondents (20%) avoid AI because they distrust its results. Among Gen Z, 35% said using AI at work feels dishonest, and 26% worried that colleagues would view them negatively for using it. On Imperial's report page, 85% of employees want further AI-related training in the next two years.
That last figure is the practical one. People are not asking to be told off. They are asking to be taught.
Why hidden use is not simply rule-breaking
If a fifth of people in the survey distrust AI enough to avoid it, and a large share of younger staff worry that using it looks dishonest, a crackdown will not create good practice. It will push work further out of sight, or it will leave people stuck doing things the slow way while their colleagues quietly use tools the organisation has never approved.
Shadow AI in the workplace is also a data-protection problem hiding in plain sight. If someone pastes customer or employee information into a consumer chatbot because that is the only tool they know, the risk is not only that the work is unofficial. It is that nobody has shown them an approved tool, a documented workflow, or a human check for the bits that matter.
The fix is not a longer ban list. It is a clear policy, a sanctioned way to work, and the skill to build automations that can be handed over, monitored and explained.
What good looks like
A team that has this under control can usually point to three things.
First, they know where AI is already being used, including the tools nobody has asked about yet. Second, they have a written AI policy people can actually follow: what is approved, what needs a sign-off, and what must not be pasted into a tool. Third, named people can turn everyday processes into approved, documented workflows, with a human in the loop where a mistake would be hard to undo.
That is ordinary operational discipline, applied to a new set of tools. It is also teachable.
How employers can close the gap
This is the work we do with employers every week, and it maps cleanly onto three TESS routes.
The AI & Automation Practitioner Level 4 apprenticeship (ST1512) is the levy or government-funded route for people who will build the workflows. It is live online, needs no coding, and takes around 15 months. In month 1, AI Foundations includes a shadow AI audit of the apprentice's own team, both the approved tools and the ones people use quietly, and a working acceptable-use position a data protection officer can read. Later modules cover process discovery, documented automations that someone else can maintain, human-in-the-loop agents, and the guardrails, monitoring and rollout plan that keep that work safe.
For leaders who need to set the rules, the Level 5 unit AI Adoption, Procurement and Governance (AU0010) is the short, Skills England-approved option. It runs over 30 hours across two weeks, with a skills test and no exam, and is levy or government funded for eligible employers. People leave with practical governance artefacts, including a policy and risk position the organisation can use, rather than a theory essay. It sits alongside the other AI Leadership units.
If you need a shorter start, our AI Ethics and Governance workshop and AI Policy Drafting workshop help a leadership, risk, legal or HR group leave with a draft position and a policy outline they can take back to counsel and the board. The workshops are a practical first step. They are not a substitute for the funded Level 4 or AU0010 routes when you need that capability in the team.
The bottom line
Shadow AI in the workplace is what you get when skill, policy and official tools lag behind the work people are already trying to do. The Imperial figures are a prompt to find that hidden use, write the rules clearly, and train people to work in the open.
- Next AI & Automation Practitioner Level 4 cohorts: 20 November 2026, 14 December 2026 and 25 January 2027, live online.
TESS Group has been delivering apprenticeships since 2009. Across our apprenticeships, 72% of apprentices achieve a distinction (results from 2024 to 2026), and we're rated 4.9/5 from 793+ reviews (Google and Trustpilot combined).
Find out more about the AI & Automation Practitioner Level 4 or book a complimentary 30-minute discovery call to talk through how a shadow AI audit, AU0010, or a Level 4 cohort would suit your team.
Sources: Imperial College London Press Office, "Imperial College London launches new division to address UK AI skills gap", 11 November 2026; Imperial News, "Imperial Lifelong Learning launches to the public with new report on AI and skills", 12 November 2026; Imperial Lifelong Learning, "What do UK workers really think about AI?" / Everyone but me: What UK employees told us about AI and skills (Censuswide fieldwork 4 to 15 September 2026).