ChatGPT Team vs Enterprise vs free: which one protects client data?

Paying for ChatGPT Team or Enterprise changes what OpenAI does with your data. It does not change what your staff paste in. Business plans keep conversations out of model training by default and add admin controls. Free and personal accounts give you neither. But no plan stops someone pasting a client's bank details into the chat box. So upgrading fixes the supplier side. The people side stays open.

Three kinds of account

Most firms have a mix. Some staff use free accounts. Some pay for Plus themselves. A few use a company plan. Each works differently.

• Free and Plus. Personal accounts. The person owns them, not the firm.
• Team (now often called Business). A shared workspace for a small group, paid per seat.
• Enterprise. For larger organisations, with more controls and a contract to match.

Names and features change. Check OpenAI's own pages before you buy. What follows is the general shape.

Training use: the main difference

On free and Plus accounts, chats can be used to improve OpenAI's models. There is a setting to turn this off. But each person controls their own setting. You cannot check it.

On Team and Enterprise, OpenAI says business data is not used for training by default. That is set for the whole workspace. Not person by person.

This is the biggest reason firms upgrade. It is a real improvement.

Admin controls on Team and Enterprise

A company plan gives you a workspace you own. In general terms, that means:

• You add and remove people. A leaver loses access.
• Chats sit in the firm's workspace, not a personal account.
• Billing is central.
• A data processing agreement is usually available, which helps with UK GDPR.

Enterprise tends to add more. Single sign-on, longer audit logs, retention settings and tighter contract terms are typical. Team is lighter on these. If your auditors or insurers want detailed logs, compare the two carefully.

Team vs Enterprise for data privacy, side by side

That last row is the same across every plan.

What no plan does

Picture Ashcombe, a small accounts firm. They buy ChatGPT Team. Training is off. Everyone uses the workspace.

Priya is chasing a late payment. She pastes a client's sort code, account number and invoice into ChatGPT. She asks for a polite reminder email. The plan lets it through. That is what it is built to do.

The data is now on OpenAI's servers. Not used for training, but stored. Whether that breaches your client agreement or your professional rules is a separate question. The plan does not answer it.

No ChatGPT plan, at any price, does these things:

• Checks a paste for card numbers, bank details or National Insurance numbers.
• Warns the person before they send.
• Stops a spreadsheet of client names being uploaded.
• Covers other AI tools. Tom may still use Gemini or DeepSeek on the side.
• Records that each person agreed your AI policy.

The other AI tools problem

Buying Team covers ChatGPT. Staff use more than ChatGPT. Claude, Gemini, Perplexity and DeepSeek are a click away. Each has its own terms. Some are much looser. Our piece on what happens to data pasted into DeepSeek covers one example.

A company plan on one tool does not reach the others. You still need a policy, and a way to see which tools are in use.

What upgrading buys you

Upgrading is still worth it for most firms that use ChatGPT often. You get:

• Training off, set centrally.
• Control over who has access.
• A contract you can point clients and insurers to.

For regulated work, that is close to a minimum. Our guide to accountants using ChatGPT with client data covers the professional side in more detail.

Closing the gap

The gap is at the keyboard. Three things help.

1. A short AI policy. Plain rules on what never goes in. Bank details, NI numbers, client names with financial detail.
2. Training. Show people what a safe prompt looks like. "Write a payment reminder for Client A" works as well as the real name.
3. A check before anything is sent. Something on the PC that spots sensitive data and takes it out.

That last step is what Helio Guard does. It checks each paste, message and upload on AI sites in Chrome and Edge. Card numbers, UK bank details and similar are swapped for labels. The question still goes through. Client names can become a stand-in such as "Client A", turned back when the answer is copied. Checked on the PC, with fixed rules. We never see what your team types. It works alongside ChatGPT Team or Enterprise, not instead of it. It is Windows only.

It is £1 a person a month, from 5 people, with a 14-day free trial and no card needed. The pricing page has the details.

Common questions about ChatGPT business plans and client data

Is ChatGPT Team safe for client data?

ChatGPT Team keeps business data out of model training by default and gives the firm control over who has access, which makes it far safer than free accounts. ChatGPT Team does not stop a member of staff pasting a client's bank details or personal data. Whether that data should go in at all depends on your client agreements and professional rules.

Does turning off chat history on a free account make it as safe as Team?

Turning off training on a free ChatGPT account helps, but the setting belongs to the individual and the firm cannot check or enforce it. A free account also has no data processing agreement with your firm and no way to remove access when someone leaves. A company workspace gives you those controls centrally.

Do I need Enterprise rather than Team for UK GDPR?

UK GDPR does not name a specific ChatGPT plan. What matters is having a lawful basis, a data processing agreement and suitable controls. ChatGPT Team can meet many small firms' needs. Enterprise adds features such as single sign-on and deeper audit logs, which larger firms or strict auditors may expect.

Related: Recruiters and ChatGPT: is it safe to paste CVs into AI?, How to redact sensitive data before pasting into ChatGPT, A client asked about your AI controls: how to reply the same day