ChatGPT vs Copilot for business: which should a UK firm use?

In the ChatGPT vs Copilot question, the deciding factor is usually where your work already lives. If your team runs on Microsoft 365, with email in Outlook, meetings in Teams and files in SharePoint, Copilot works inside those tools and follows the permissions you already have. If your work is spread across other tools, or you mainly want a flexible assistant for writing, analysis and research, ChatGPT is often the simpler fit. Either way, use a business plan rather than personal accounts, and give staff clear rules.

What each one is

ChatGPT is OpenAI’s AI assistant. You use it in a browser or app: you type or talk, upload files, and it drafts, summarises, analyses and answers. There are personal plans for individuals and business plans for organisations, currently ChatGPT Business (formerly ChatGPT Team) and ChatGPT Enterprise.

Copilot is Microsoft’s name for several different products, which causes most of the confusion:

• Copilot Chat. An AI chat included with eligible Microsoft 365 business subscriptions. It works mainly from the web and from whatever you paste, upload or have open, and in Outlook and Teams it can work with the item you have open. What it can do inside Word, Excel and PowerPoint depends on your Microsoft 365 licence.
• The paid Copilot licence. An add-on to Microsoft 365, sold as Microsoft 365 Copilot (the app itself is now called Microsoft Copilot). It works across the emails, meetings, chats, files and SharePoint sites each person already has access to, inside the Microsoft 365 apps.
• Copilot on a personal account. The consumer version people use at home. Microsoft’s enterprise data protection applies when people sign in with their work account, so staff should use that account for anything to do with work.

Business plans and your data

Here is what each provider’s own documentation says at the time of writing:

• ChatGPT Business and Enterprise. OpenAI says it does not use data from these plans, including what you type and what comes back, to train its models by default, and that your organisation owns its inputs and outputs. Admins control who has access.
• Personal ChatGPT plans. Conversations can be used to improve OpenAI’s models unless the user switches that off in their settings. Fine for personal use, not for client work.
• Copilot with a work account. Microsoft says prompts, responses and the data Copilot reaches through Microsoft 365 aren’t used to train its foundation models, and are handled under the same commitments as the rest of your Microsoft 365 content. Copilot only shows each person content they already have at least view access to.

Two cautions. Your data may be processed outside the UK: Microsoft’s documentation says Copilot requests from outside the EU may be processed in the US, the EU or other regions, and OpenAI’s data residency options vary by plan. If location matters to you or your clients, check the current terms for your plan before you sign. And terms change, so read them yourself rather than relying on a blog post, including this one. This is practical guidance, not legal advice. For what staff should and shouldn’t share with either tool, see is it safe to use ChatGPT with customer data?

Where Copilot fits

Copilot makes most sense when your business already runs on Microsoft 365 and the information people need sits in Outlook, Teams, OneDrive and SharePoint. With the paid licence, it can work across all of that at once: pull together an email thread, the meeting where it was discussed and the files behind it, without anyone copying data into another tool. Your existing Microsoft 365 permissions, retention policies and sensitivity labels still apply.

The catch is permissions. Copilot shows people what they are already allowed to see, which is only safe if your permissions are right. In many businesses, a SharePoint site or shared folder was opened to everyone years ago for convenience. Before Copilot, nobody stumbled on the salary spreadsheet in an old shared folder. With Copilot, an innocent question can bring it up. Check who can see what before you switch it on. If nobody in your business owns your SharePoint permissions, don’t buy the paid licence until someone does.

If you already have Microsoft 365, Copilot Chat is a sensible place to start before buying the paid licence. It gives staff a work-safe assistant under their work account, and it shows you whether people will use one at all.

Where ChatGPT fits

ChatGPT suits businesses that don’t run on Microsoft 365, and teams whose work is mostly thinking on the page: drafting proposals, rewriting for different audiences, analysing a spreadsheet, preparing for a meeting or working through a problem step by step.

ChatGPT business plans can also connect to tools such as SharePoint, Outlook and Teams, and OpenAI says those connectors only see what each person is allowed to access. The same warning applies: get permissions right first, and let an admin decide which connections are switched on.

Do you need both?

Usually not, at least at first. Two tools means two sets of settings, two places your data goes, two sets of rules for staff and two bills. For most small businesses the better route is to choose one, use it well, and revisit the choice once you know what people actually use it for.

There are sensible exceptions. A business on Microsoft 365 might give the paid Copilot licence to the people who spend their day in Outlook and Teams, and ChatGPT Business to a small team that does heavy drafting or analysis. That works, provided each tool is approved, set up on business accounts and covered by the same rules. What doesn’t work is an official tool nobody uses while the real work happens on personal accounts.

A two-week trial plan for ChatGPT vs Copilot

Don’t decide from demos. Test both, or the one you are leaning towards, on your own work for two weeks.

1. Set up properly first. Business accounts only, signed in with work email, with the data and sharing settings checked by whoever runs your IT.
2. Pick real tasks. Choose a handful of jobs your team does every week, such as summarising a long email thread, drafting a client letter from notes, turning meeting notes into actions, or answering a question from your own policies.
3. Pick a small, mixed group. Include someone keen, someone sceptical and someone who handles sensitive information.
4. Week one: the same tasks in each tool. Note how long each took, how much editing the output needed and anything it got wrong.
5. Week two: your own data. Try the tasks that depend on your files, inbox and meetings. This is where the real differences show, and where any permission problems appear.
6. Decide on evidence. Compare your notes, not impressions. Choose one tool, the people who will use it and the tasks it is approved for, then write the rules down.

The rules staff need either way

At Alter Domus, where I was Head of AI & Automation, the internal assistant domusAI grew from a 605-person beta to more than 4,000 people using it every day (more on the Alter Domus work). In my experience, clear rules and practice on real work matter as much as the choice of tool. Whichever you choose, staff need to know:

• Which tools are approved. Work accounts only. No personal accounts for client or company information.
• What never goes in. Passwords, bank details, health information, full customer lists and anything under a confidentiality agreement, unless there is a specific, reviewed reason.
• That a person checks the output. Both tools can be confidently wrong. Anything sent to a client or used for a decision is reviewed first.
• What to do after a mistake. Tell a named person straight away, without fear of blame.

You can start from the free AI acceptable use policy template, which covers ChatGPT, Copilot and other tools. Rules stick better with practice, which is what AI training for teams is for: sessions built around your own tasks, in the tool you have chosen.

If you want an independent view on which tool suits your business, or help running the trial, book a free 30-minute AI consultation. We will look at how your team works today and where an AI assistant would actually save time.