Client login
AI Agents
Services
AI Property AcceleratorAI Hub: guides and articlesPricingAboutContact
Book an AI Audit: £197Free Brain Score
Safety, GDPR and policy

What Is Shadow AI? A Plain Guide for Small UK Firms

Shadow AI is staff using AI tools the business has not approved. Here is why it happens and how to fix it calmly.

By Imz, Founder · AI Solutions Architect5 min read12 November 2026
Isometric drawing: a grid of approved tools with three unapproved ones in black

Shadow AI is when staff use AI tools that the business has not approved, or does not even know about. It might be a free chatbot on a personal phone, or an AI add-on someone switched on in a browser.

Most of the time nobody is being sneaky. They are trying to get through a busy day. But if you run a small firm, it is worth knowing what is happening, because the risks land on you. This article offers general guidance only and is not legal advice.

What is shadow AI?

The name comes from "shadow IT", which means technology used at work without the business knowing. Shadow AI is the same idea with AI tools.

Here are some everyday examples:

  • A fee earner pastes a client email into a free chatbot to get a reply drafted.
  • An office manager uploads a spreadsheet of customer details to get a summary.
  • Someone installs an AI note-taker on their own account to record client calls.
  • A team member uses an AI browser extension that reads every page they open.

The common thread is that the business has no list of what is used, no rules, and no record.

Why shadow AI happens

People use these tools because they work. They save time on writing, summarising and sorting. If the business has not offered a safe option, staff will find their own.

There is some evidence this is widespread. Microsoft and LinkedIn's 2024 Work Trend Index reported that 78% of AI users were bringing their own AI tools to work. The figure was 80% at small and medium-sized companies.

Treat that number with care. It comes from a global survey of 31,000 knowledge workers in 31 markets, run by Edelman Data & Intelligence in early 2024. It is not a UK figure and it is not recent. It does show the direction of travel.

The lesson is simple. A ban rarely works. A safe, approved route usually does.

The real risks for a small UK firm

Client data in free tools

This is the big one. The NCSC warns people not to put sensitive information into public AI tools. It says your queries are visible to the provider, are stored, and will likely be used to develop the service.

If that text holds personal data, you are the one responsible under UK GDPR.

No record and no control

If you do not know a tool is in use, you cannot check its terms. You cannot tell what was shared. You cannot remove access when someone leaves.

The ICO's accountability guidance for AI says overall accountability stays with your organisation, even when you buy or outsource an AI tool. It also suggests choosing another solution if you cannot tell whether a third-party tool is compliant. You cannot do that for a tool you have never heard of.

Wrong answers sent to clients

AI can sound confident and be wrong. It can invent a clause, a date or a figure. If a member of staff sends that on without checking, it goes out under your firm's name.

Mixed-up work and personal accounts

Chat history on a personal account stays with the person, not the firm. When they move on, so does the record of what they shared.

How to spot it without a witch hunt

Do not start by hunting for culprits. If people fear punishment, they will hide their use and the risk gets worse.

Try these gentler ways to find out:

  • Ask in a team meeting: "What AI tools do you use, and what for?" Say clearly there is no blame.
  • Run a short, anonymous survey with the same question.
  • Ask your IT provider which AI apps and browser extensions are installed on work devices.
  • Look at expense claims and card statements for AI subscriptions.
  • Check which apps staff have connected to your email and Drive.

Thank people for being honest. You want the true picture more than you want a clean record.

A calm 5-step fix

  1. Ask what people use. Make a simple list: tool, who uses it, what for, and what data goes in.
  2. Approve a short list. Pick a few tools you are happy with and say no to the rest, with reasons. Keep the list short enough to remember.
  3. Give staff a safe, paid tool. Business plans usually come with better terms than free accounts. Free tools stay popular if nothing else is on offer.
  4. Write a one-page policy. Say what is approved, what must never go in, and who to ask. Our free AI policy template for UK businesses gives you a start.
  5. Review every quarter. New tools appear all the time. Every three months, ask again, update the list and remind people of the rules.

Training helps too. People follow rules better when they understand the reasons. See AI training for teams.

What to do when you find something

Stay calm and be practical. First, find out what was shared. If client or personal data went into a free tool, note the tool, the date and roughly what was entered.

Then decide whether it needs reporting. A personal data breach can have reporting duties, so ask a data protection adviser if you are unsure. Move the work to an approved tool and close the personal account where you can.

Finally, use it as a lesson for the whole team. Share what you learned and what has changed, without naming anyone.

What goes in an AI usage policy

An AI usage policy, sometimes called an AI acceptable use policy, can be one page. It does not need legal language. Aim to cover these points in plain words:

  • Which tools are approved, and how to ask for a new one.
  • What must never be typed in, such as client names, health or financial details, passwords and unpublished deals.
  • That a person must check AI output before it goes to a client.
  • That staff use work accounts only, not personal ones.
  • What to do if something goes wrong, and who to tell straight away.

The ICO's eight questions on generative AI are a useful check before you finalise it. They cover your lawful basis, DPIAs, transparency and security. If personal data is involved, ask a data protection adviser or solicitor to look it over.

When AI is the wrong tool

Some jobs should not go through a general AI tool. Examples include sensitive client files, legal advice, and decisions about people. If in doubt, leave it out and ask.

Where Agentic Scale fits

The cure for shadow AI is a better official option. That is the part we build.

Our AI Agents give staff approved AI employees that already know the firm's files, so the quick, tempting tool and the permitted tool become the same thing. Every send still needs a human yes.

An AI Audit includes a frank list of what your team is using now. It costs £197, which is credited in full against a build within 30 days. AI Shield then repeats the check on a schedule. The policy and the habits remain yours to keep up.

Sources

Want this working in your business? The AI Audit maps where AI fits and prices it. £197, credited if you build.

FAQ

Questions,answered.

Anything else, ask Imz directly.

What is shadow AI?
Shadow AI is when staff use AI tools that the business has not approved or does not know about. Examples include pasting client emails into a free chatbot or using an AI note-taker on a personal account. Usually the aim is to save time.
Why is shadow AI a risk?
Staff may paste client or personal data into free tools, which the NCSC advises against. The business then has no record, no control over the terms, and no way to check what was shared or what was sent on.
Should I ban AI tools at work?
A blanket ban rarely works because people will keep using their own tools quietly. A better route is to find out what is used, approve a short list, provide a safe paid tool, and write a one-page policy.
What should an AI acceptable use policy include?
Keep it to one page. List approved tools, what must never be typed in, the need for a person to check output before it reaches a client, work accounts only, and who to tell if something goes wrong.
Taking new clients for November

Not sure where to start?Book an AI Audit.

We look at how you work today and tell you honestly what AI should and shouldn't do for you. £197, credited in full if you build within 30 days.