
Illustrative: a man takes euro notes out of his wallet, photographed in Croatia in January 2023. Image: European Commission (Audiovisual Service) / Wikimedia Commons, CC BY 4.0, cropped
British workers are spending an estimated £958 million (about $1.2 billion) a year of their own money on generative AI tools they use for work, and nearly a third of AI users are doing it without their employer knowing, according to a survey of 25,000 people by Deloitte UK. The accounting firm calls it the largest study of workplace AI use carried out in a single country.
One in six AI users pays for it themselves
Deloitte’s first GenAI Workforce Survey, carried out for the firm by Ipsos UK between May 7 and June 10, found that 63% of working adults aged 18 to 70 knowingly use generative AI for work. Almost half (46%) say they use free tools, 34% use outside tools their employer pays for, and 17% use tools built in-house.
The money comes from the 17% of AI users who pay for at least one AI tool themselves to use for their job. Deloitte scaled that up to its £958 million estimate of what workers spend each year. With top plans for ChatGPT and Claude costing up to $200 a month, a personal subscription is not a small outlay. The figures are Deloitte’s own and haven’t been independently verified.
“They don’t want to wait for permission”
The survey suggests many staff are moving faster than their employers. Some 31% of AI users say they use it without their employer’s knowledge, often called shadow AI, and bring their own tools to work. Almost a quarter (24%) of all workers now use AI every day for their job, and 65% of AI users say their organisation lacks convincing leadership on how it should be used.
Hayley McKelvey, Deloitte UK’s chief AI officer, said the demand is already there:
UK workers are showing they don’t want to wait for permission to use GenAI. Many are already using free tools or pay for premium versions themselves to help them get work done.
Hayley McKelvey, chief AI officer, Deloitte UK
Shadow AI is a real worry for employers: company documents pasted into a personal account sit outside the firm’s security controls, and personal logins and API keys are a target in their own right, with stolen Claude, Gemini and ChatGPT access now sold cheaply on the dark web.
Time saved, but a stigma too
Workers say AI saves them an average of 70 minutes a week, and most of that time goes into doing more work for the same employer. Yet 23% think there is a stigma attached to using AI at work, and 64% of weekly users worry their managers will conclude AI can do their jobs. Deloitte says people who feel that stigma are more likely to hide their use.
For most people, the uses are still basic. The top three are searching for information (43%), drafting emails (43%) and writing summaries (31%). About half of AI users say they have had no formal training on using it safely and well.
Paul Lee, Deloitte UK’s head of industry insight, said that was the real finding:
The story here isn’t that workers are using GenAI, it’s that they are using it despite limited training, patchy guidance and, in some cases, without their employer’s knowledge.
Paul Lee, partner and head of industry insight, Deloitte UK
Big software firms are pitching exactly this gap to employers, with Microsoft selling its new Copilot as an “OS for work”. Deloitte, which sells AI consulting to businesses, says it will repeat the survey about every six months.
Why it matters
When workers are paying out of their own pockets for tools they use at work, and hiding it, employers have lost track of how AI is being used at their own companies. The firms that give staff approved tools, training and clear rules will get both the productivity and the security; the rest are relying on personal subscriptions they can’t see.
Sources: Deloitte UK press release (September 16, 2026), Deloitte UK GenAI Workforce Survey.


