Is ChatGPT Private? What It Keeps and What to Switch Off

If you've ever pasted a client email, a contract or a spreadsheet of customer details into ChatGPT and then wondered where it went, you're not alone. It's one of the most common worries I hear from business owners, and it's a sensible one. The good news is that OpenAI publishes clear answers, and the settings that matter take a couple of minutes to change.

Here's what happens to your chats, what you can switch off, and the simple rules I'd follow before you paste in anything to do with a client.

Is ChatGPT private?

Not by default, in the sense most people mean. On personal accounts, OpenAI says it may use what you submit, including prompts, responses, images and files, to improve its models, depending on your settings. You can turn that off. OpenAI also says that a limited number of authorised staff and trusted service providers may access content when needed, for example to investigate abuse or a security incident, to give you support, for legal matters, or to improve model performance unless you've opted out.

OpenAI's own advice is blunt: don't enter sensitive information that you wouldn't want reviewed or used. That's the line I'd keep in your head.

Does ChatGPT use your chats for training?

On personal accounts it can, unless you turn off the setting called Improve the model for everyone. When it's off, OpenAI says your new conversations won't be used to train its models. When you're signed in, the setting applies to your account across your devices.

Two details catch people out. Turning the setting off doesn't delete your existing chats, and archiving a chat doesn't change whether it can be used for training. And if you give feedback on a response, such as a thumbs up or thumbs down, OpenAI says the entire conversation linked to that feedback may be used to train its models, even after you've opted out.

How do I stop ChatGPT using my data?

On the web, when you're signed in:

1. Open your account menu and select Settings.
2. Select Data controls.
3. Select Improve the model for everyone, turn it off, and select Done.

On iOS and Android, open the sidebar, select your profile icon to open Settings, choose Data controls and turn off Improve the model for everyone. If you use ChatGPT without signing in, you can turn the same setting off in Settings, and your choice is saved in that browser.

What is a temporary chat in ChatGPT?

If you only remember one thing from this guide, make it this: the settings are yours to change, and most people never open them. It takes two minutes, and it changes what happens to every chat you have from then on. Do it on every device and every account your team uses for work, including the phone app.

A temporary chat is the closest thing to a private conversation on a personal account. While it stays temporary, OpenAI says it doesn't appear in your chat history, doesn't create or update memories and isn't used to improve its models. OpenAI may keep a copy for up to 30 days for safety purposes.

Before the conversation starts, you can choose whether it's personalised. A personalised temporary chat can use your existing memories and custom instructions; an unpersonalised one doesn't. If you save a temporary chat, it becomes a normal chat and follows your usual settings.

Does ChatGPT keep deleted chats?

You can delete chats one at a time or clear your whole history. OpenAI says cleared chats are deleted from its systems within 30 days, unless they've already been de-identified and disconnected from your account, or it has to keep them for security or legal reasons. Turning off training doesn't delete anything, so if you want a chat gone, delete it.

Is ChatGPT safe for client data?

This is where the type of account matters. OpenAI says that, by default, it doesn't use content from ChatGPT Business, Enterprise, Edu or ChatGPT for Healthcare workspaces to train its models, and those workspaces can have their own retention and access settings. So if your team regularly works with client material, a managed business workspace gives you stricter defaults than a personal account.

Even then, I'd keep a few house rules:

• Strip out names and identifiers before you paste. Replace them with Client A or Supplier B.
• Never paste passwords, bank details, health information or anything covered by a confidentiality agreement.
• Use temporary chats for one-off sensitive jobs on a personal account.
• Check what your team is using. Personal accounts used for work are where most leaks start.
• Write it down. A one-page AI policy saves arguments later.

If you're not sure what your staff are already pasting into AI tools, my piece on what your team is doing with ChatGPT behind your back is worth five minutes.

A prompt to clean client details before you paste

Here's a prompt I like for preparing material. Run it on text you've already stripped of the obvious details, as a second check, not as your only check.

Copy this prompt

Read the text below and list every detail that could identify a person or a business: names, email addresses, phone numbers, addresses, account or order numbers and anything similar.
Then rewrite the text with each one replaced by a neutral label such as [Client A], [Email 1] or [Address 1].
Don't change anything else.
Text: [paste here]

And one for your team, so everyone checks before they share:

Copy and paste this prompt

I'm about to ask an AI assistant for help with this task: [describe the task].
Before I paste anything, list what I should remove or anonymise first, and suggest how to describe the task without sharing confidential details.

How do I see what ChatGPT knows about me?

There are three places to look. Memory, which controls how ChatGPT personalises its answers, lives in Settings, Personalization, and you can view, edit or delete memories there. Custom instructions are in the same place. And if you're on a Free, Go, Plus or Pro account, you can request an export of your data from your settings. On Business and Enterprise workspaces, exports go through the workspace owner.

OpenAI also states that it doesn't sell your data or share your ChatGPT conversations with advertisers.

Tools that help you keep client data safe

A few everyday tools make the habits above easier: a password manager such as 1Password or Bitwarden so no one ever pastes a login into a chat, and your own documents kept in Google Workspace or Microsoft 365 with sharing set carefully, so AI tools only ever see copies you've chosen to share. If your website collects any of this data, check you've got a privacy policy on your site too.

If you'd like a simple AI policy and settings checklist written for your team, that's the kind of work I do with business owners.

Frequently asked questions

Is ChatGPT private?

Not fully by default. On personal accounts OpenAI may use your chats to improve its models unless you turn off Improve the model for everyone, and authorised staff may access content in limited cases. Temporary chats and business workspaces give you stricter defaults.

How do I stop ChatGPT from training on my chats?

Go to Settings, Data controls and turn off Improve the model for everyone. When you're signed in, it applies across your devices. Feedback such as a thumbs up can still send that whole conversation for training.

Are ChatGPT temporary chats private?

Temporary chats don't appear in your history, don't create memories and aren't used to improve models, but OpenAI may keep a copy for up to 30 days for safety purposes.

Does ChatGPT use business workspace data for training?

OpenAI says that, by default, it doesn't use content from ChatGPT Business, Enterprise, Edu or ChatGPT for Healthcare workspaces to train its models.

Can I paste client information into ChatGPT?

Only after anonymising it and checking your settings. Remove names, contact details and account numbers, never paste passwords or confidential material, and use a business workspace or temporary chats for sensitive work.