British workers are spending nearly £1 billion of their own money on AI tools they use for work.
That is Deloitte's estimate from a survey of 25,000 UK workers. One in six GenAI users pay for at least one tool of their own pocket and about a third use GenAI without their employer's knowledge. (Deloitte press release, 16 Sep 2026)
It is easy to see why this is happening. When someone uses ChatGPT in their personal life, then steps into a work environment with either no AI tools or a poorly implemented Microsoft Copilot that is incapable of making a simple edit to a deck, it feels like stepping back in time.
But the risks around data, privacy, and security are very real.
Adoption without depth
Deloitte's core finding is the one that should make every leadership team sit up: GenAI use is widespread, but still shallow.
Sixty-three percent of working adults now use GenAI for work. Nearly a quarter use it every day. And yet the top three use cases are searching for information, drafting emails, and creating summaries. These are simple familiar tasks, not a rethink of how work gets done.
Simply put, employees are bringing tools to work that meaningfully increase risk and exposure, but applying those tools to the most basic of tasks.
Perhaps this is due to the fact that only 20% of AI users would class themselves as highly proficient, and nearly half of employees who use AI at work have had no formal training. Only one in five say the training they did get was mandatory. When organisations do mandate training, it is mostly digital and one-shot. As one HR leader told me; "People had this tool but they didn't have the time to go to the webinars, and you weren't giving people the airspace to actually try and utilise their new skills."
If we're not investing in our people's skills, or investing in making space in their day to help them practise them, is it any wonder that we see such shallow usage of AI?
ShadowAI is rife
Much of this usage is hidden from employers. Nearly a third say they use the tools without their employer's knowledge. Half of this usage is on a free tool. That is a massive concern for a business. To an uninformed user, it might feel fine to use the free version of ChatGPT to rewrite an email, but as a business leader, you have no idea what data is being shared and the exposure you open your business to.
Hayley McKelvey, chief AI officer at Deloitte UK: "Workers who feel a stigma around adopting GenAI tools are more likely to conceal their use."
So if a leader is trying to control AI usage with by blocking it entirely people find other ways of using it - as we're already seeing. The alternative is to give people access in a structured way and give them permission to experiment. That means giving good guidance, setting guardrails, being honest that no one has the perfect answer to any of this, and allowing for time to experiment. The more forward-thinking organisations are taking exactly this approach.
Paul Lee, partner and head of industry insight at Deloitte UK, made the same point from another angle: the story is not that workers are using GenAI. It is that they are using it despite limited training, patchy guidance, and, in some cases, without their employer's knowledge.
The difference between an AI policy and a manifesto
Again, according to a Deloitte report, fewer than half of GenAI users say their organisation even has a GenAI policy. Only 28% say there is a policy they can find, understand, and trust as up to date. Sixty-five percent do not hear leaders talk about GenAI with a clear understanding of it.
This is the part that I find pretty concerning, and I've seen this myself, is leaders of organisations that have got excited about using ChatGPT on the weekend and come into the office on a Monday excitedly saying, "We need to use AI!", and not really knowing what they mean.
AI is not a tool. It's not something you just wake up and decide that people need to start using. AI is a technology. That means we need to think deeper about how it impacts our business. This is not something leaders can delegate away, they need to sit with this and truly understand the implications of AI, both in terms of the positive and negative impact on the business.
The first step to solving this is leadership creating, as Erica Farmer put it in our last webinar, an AI manifesto.
The manifesto sets the expectation for how an organisation will use AI. Making people feel psychologically safe, giving them room to experiment, and giving them the support they need to understand how they can safely and productively apply AI to their work.
Without that, people do what Deloitte is measuring in the wild. They talk to friends in other businesses, they bring their own tools, they hide the use when they feel stigma. Shadow AI is what you get when manifesto and capability lag behind usage.
Would be very interested to hear if any of this resonates and it's a pattern that you've seen in your organisations.
Thanks, as always, Andy
Sources
- Deloitte UK GenAI Workforce Survey 2026 (25,000 UK workers, fieldwork 7 May to 10 June 2026): https://www.deloitte.com/uk/en/issues/generative-ai/genai-workforce-survey.html
- Press release (nearly £1bn / £958m personal spend / shadow AI): https://www.deloitte.com/uk/en/about/press-room/british-workers-spend-one-billion-pounds-of-their-own-money-on-gen-ai-for-work.html
- Leadership & training deep dive: https://www.deloitte.com/uk/en/issues/generative-ai/genai-workforce-survey-leadership-training.html
