ChatGPT at work: what you as an employee may and may not do

An article for you, not about you

Most articles about AI at work are written for managers, HR people or compliance officers. For you, the employee opening ChatGPT at four in the afternoon with three emails left to write and zero energy, clear guidance is rarely available. This is that article. Not a cautionary tale, not a ban, not a "think carefully before you type anything". Just an honest look at what you may do, what you may not, and how to make the right call for yourself, even when your employer has no policy about it.

First, the reality

ChatGPT has quickly grown into the best-known and most-used general AI chatbot at work, and in almost no organisation is that neatly arranged. Studies show that in knowledge work and office roles a majority now use AI tools: depending on sector, country and method of measurement, the figures range from well over half to about three-quarters. And in a large share of those cases it happens without formal permission from the employer, without a policy, and without clear agreements.

We call that shadow AI, meaning AI use that stays under the radar. And the first person to suffer if it goes wrong is you, not the leadership. That is why it matters that you know the rules yourself, even if nobody has explained them to you.

The three questions you should always ask yourself

Before you type anything into ChatGPT (or another AI tool), three short questions help you decide whether it is okay:

Question 1: Would I comfortably share this with a stranger on a café terrace? Because technically that is more or less what it comes down to. What you type into a personal ChatGPT account goes to OpenAI's servers and, unless model training is disabled in the data controls, may be used to improve OpenAI's models. With business services such as ChatGPT Business, ChatGPT Enterprise and Microsoft 365 Copilot, the providers state that prompts and responses are not used to train foundation models by default, subject to the plan, settings and contract. If the answer to question 1 is "no", do not type it into an AI tool either, certainly not a personal account.

Question 2: Is this information my own, or someone else's? An email you wrote yourself and want to improve, your information. A client conversation you took notes on, someone else's information. An application letter from a candidate, someone else's information. A patient record, very much someone else's information. When it comes to your own work and text, the threshold is low. However, for third-party information, it quickly becomes much more complicated, and often the answer is simply: do not.

Question 3: How would I feel if my employer, client or colleague looked over my shoulder? Not a moralising sermon, but a practical question. If the answer lies somewhere between "uncomfortable" and "panic", that is a sign you had better handle it differently.

What is usually safe

Where you should think twice

What you should never do

What if your employer has no policy?

That is the situation for the majority of people reading this. And yes, it is not your responsibility to write that policy. But there are three things you can do yourself:

You can raise it with your manager or HR, not to criticise, but as a practical question: "I sometimes use ChatGPT to improve my emails, is there a line on that?" Many managers are more open to it than you think, because they use it themselves and feel uncomfortable about it.

You can set yourself a short rule: nothing goes in that I would not also discuss in a lift with a stranger. That covers 90 per cent of the risks.

And you can watch whether your employer offers AI literacy training. Since 2 February 2025, that has been a legal obligation for the employer under Article 4 of the EU AI Act: so that obligation already applies, not just somewhere in the future. As an employee, you may reasonably expect your employer to provide clear guidance and appropriate AI literacy training where you use AI systems at work. If you want to know what the EU AI Act expects from employers, you will find it in a separate insight.

In closing

AI at work is not a luxury problem and not an elite debate. It is in your browser, it makes your working day shorter, and it can also get your career into trouble if you use it the wrong way. The good news is: you do not have to become a lawyer to do it well. A few simple questions, a healthy sense of what belongs to whom, and the willingness to be the first to raise the subject, that is all it takes to stay on the right side.

And to your employer: if you are reading this article and thinking "we have not actually arranged anything about this", that is exactly what AIAdopt exists for. A practical microtraining programme that brings employees, management and HR up to speed together, with an assessed certificate that supports the AI literacy built up, exactly what you need to evidence your Article 4 effort.

👉 View the base package or build your own package tailored to your organisation.