Call

Applied AI, 21 November 2026, 4 min read

ChatGPT for business: which plan to buy, and what it does with your data

Half of your staff already use ChatGPT. The difference between their plan and a business plan is what OpenAI does with what they type: the plans and prices, the data rules behind each, and three controls to put in first.

Most businesses did not decide to adopt ChatGPT. Somebody in accounts started using it to draft letters, somebody in sales used it to summarise a tender, and by the time anyone asked, half the office had an account. The question that lands with the owner is which version to pay for, and whether paying changes anything. It does, and the thing it changes is what OpenAI does with what your staff type.

The plans, in one table

OpenAI sells ChatGPT to individuals and to organisations. The two are different products with different rules.

Plan Who it is for Price before VAT Your data used to train OpenAI’s models?
Free, Go, Plus, Pro One person, on a personal account Free, or a monthly fee for the paid tiers Yes, unless the person turns it off in their settings
ChatGPT Business A business, two seats or more, managed from a workspace About £15 a seat a month paid yearly, or £18 monthly No, by default
ChatGPT Enterprise Larger organisations, sold by OpenAI’s sales team Not published; third-party reports put it around £45 a seat, with a minimum seat count No, by default

OpenAI sets its prices in US dollars and shows the pound figure at its UK checkout, so the figures above are what it showed in November 2026, before VAT, and they move without much warning; check the pricing page before you sign. A higher-usage Business seat exists at about £75 a month for people who hit the limits, and OpenAI caps Business workspaces at a few hundred seats, above which it points you at Enterprise.

What the business plans change

Training

On a personal plan, conversations are used to train OpenAI’s models unless the person has found the setting and turned it off, and most have not. On Business and Enterprise, workspace data is excluded from training by default (OpenAI, Enterprise privacy). That is the single biggest reason to pay: a client’s contract pasted into a personal account may inform a future model; pasted into a Business workspace, it will not.

Who owns the account

On a personal plan the account belongs to the member of staff, and when they leave, the history of everything they asked leaves with them. A Business workspace belongs to the company: an administrator adds and removes people, sees which tools and connectors are enabled, and can close a seat the day someone goes.

A contract you can show someone

Business and Enterprise come with OpenAI’s data processing terms, which is what your data protection adviser, your insurer and a client’s security questionnaire will ask for. A personal account has consumer terms, and no business can rely on them.

What Enterprise adds

Single sign-on through your Microsoft 365 identity, so leavers lose ChatGPT when they lose everything else; longer context and higher limits; and options on where data is stored. For most businesses of our size the honest answer is that Business is enough, and that Enterprise becomes worth it when the seat count reaches the point where OpenAI will talk to you anyway.

Not training on it is not the same as not storing it

Excluding your data from training does not mean it is not sent to OpenAI, processed on its servers and kept for a period. It is, on every plan. The business plans give you controls over retention and a contract about it; they do not make the tool private. Two consequences follow.

The first is that some things should never go in, on any plan: anything you have promised a client to keep on your own systems, anything under a confidentiality agreement that names where data may be processed, and personal data you have no basis to share with a processor in another country. Those workflows belong on AI that runs on hardware we own, where the data does not leave the North East.

The second is that the people typing need to know the rule, which is why the AI use policy comes before the licences. A policy nobody has read does not change what gets pasted.

Three controls to put in before you buy seats

  1. Find out what is in use. Before deciding on ChatGPT, see which AI tools staff already use and what goes into them. In most businesses the list is longer than anyone expects, and ChatGPT is on it alongside tools nobody has heard of. Our shadow AI protection shows the tools in use from the browser, where they are used, and lets you allow the approved ones and block the rest.
  2. Decide which tool for which job. If your staff live in Microsoft 365, Microsoft 365 Copilot works inside the documents and mailboxes they already have, within the permissions they already hold, and the data stays in your tenant. ChatGPT is better for some tasks and worse for others. A business can sensibly run both, with a line between them that people understand.
  3. Give every tool an owner. Someone in the business owns the ChatGPT workspace: who is in it, what is connected to it, and what the policy says. Without that, seats accumulate, connectors get switched on and nobody reviews it, which is the state managed AI exists to prevent.

What we do

We set businesses up on ChatGPT Business or Copilot, or both, with the discovery and policy done first and the training done for each role: what the tool is for, what never goes into it and how to check what comes out. We also run AI on our own hardware for the work that cannot leave. If the question in your business is simply “should we be paying for this”, that is a short conversation, and the service desk answers within three rings on 01642 661888.

Ask us how to put ChatGPT on a proper footing

Call and we answer within three rings. Or tell us what you need and we will call you back at any time.

01642 661888