Ask around your business and you will probably find that some of your people already use ChatGPT, Copilot or something similar. They use it to tidy up emails, summarise long documents, draft proposals or make sense of a spreadsheet. Most do it quietly, on personal accounts, because nobody has said whether it is allowed.
That is not a reason to panic. It is a sign that your team has found something useful. But it does need managing, because the way it is happening now carries risks you probably haven't agreed to.
What the real risks are
- Company and customer information leaving the business. When someone pastes a customer contract, a pricing sheet or a staff matter into a free personal account, you no longer control where that information goes or how it is used.
- Mistakes that nobody checks. AI tools are confident even when they are wrong. A figure, a clause or a date can be invented and look perfectly plausible. If the output goes straight to a customer, the mistake is yours.
- Decisions about people. Using AI to screen job applicants, assess staff or make decisions about customers' money or insurance can bring legal obligations that a casual user won't know about.
- Uneven use. A few enthusiasts get real value while everyone else does it the old way, and nobody shares what works.
Why banning it rarely works
Some businesses respond by blocking the tools. In practice, people carry on using them on their phones, and the business loses both the benefit and any visibility of what is happening. A ban also sends a message that the business isn't interested in better ways of working, which is rarely the message an owner wants to send.
The better question is not "should we allow it?" but "how do we use it well and safely?"
Five steps that make the difference
- Find out what is already in use. A short, no-blame survey: which tools, for what work, on which accounts. You will learn where the value is as well as where the risk is.
- Agree simple rules. What can never go into a public AI tool (customer details, staff matters, anything confidential), what needs checking before it is used, and who to ask when unsure. One page is enough.
- Provide the right tools, set up properly. A business account on an approved tool, with the data protection settings switched on, removes most of the temptation to use personal accounts. Many businesses already pay for AI features in software they own without realising it.
- Show people how to use it on their own work. Generic training is forgotten quickly. Short sessions built around each team's real documents and tasks stick, because people leave with something they can use the same afternoon.
- Check back after a month. Who is using it, for what, and what has changed? Cut the licences nobody uses and share what works.
Start with the rules
If you do one thing this week, agree the rules. Our free Safe AI checklist gives you a one-page starting point your team can adopt straight away: what never goes into a public AI tool, how to check what comes out, and what owners should put in place.
Where we can help
If you would like it done properly, AI Adoption for Your Team chooses the right tools for your work, sets them up safely with your IT provider, agrees the rules and trains each team on its own work, with a follow-up after 30 days to make it stick. If you would rather start by seeing what AI could do for the business as a whole, begin with an Executive AI Session.
