Guide

How to stop ChatGPT from training on your chats

On personal ChatGPT plans, your conversations can be used to improve OpenAI's models unless you say otherwise. Here's the setting that stops it, the exceptions worth knowing, and what changes on business plans.

Updated 18 October 2026

Quick answer: go to Settings → Data controls and turn off Improve the model for everyone. From then on, your new conversations won't be used to train OpenAI's models, and your chat history stays as it is. Business, Enterprise and Edu workspaces are already excluded from training by default.

These steps were checked against OpenAI's help pages on 18 October 2026. Menu labels may differ slightly by app version.

Who needs to change this

OpenAI says it may use content from its services for individuals (Free, Plus and Pro) to train its models, with an opt-out. If you use ChatGPT through an employer or school on a Business, Enterprise, Edu or ChatGPT for Healthcare workspace, OpenAI doesn't train on that content by default, so there's normally nothing for you to switch off.

Turn it off on the web

Open your account menu.
Select Settings.
Select Data controls.
Select Improve the model for everyone, switch it off, then select Done.

When you're signed in, the choice is tied to your account, so it carries over to your other devices.

Turn it off in the iPhone or Android app

Open the sidebar.
Tap your profile icon to open Settings.
Tap Data controls.
Switch off Improve the model for everyone.

The other way: OpenAI's Privacy Portal

Instead of the in-app switch, you can make the request Do not train on my content in OpenAI's Privacy Portal. OpenAI says either route is enough on its own; you don't need both. If you already opted out earlier through OpenAI Support or a privacy request, that opt-out still stands.

If you use ChatGPT without signing in

Signed-out ChatGPT has a Settings menu too, with the same Improve the model for everyone switch. The difference is that your choice is stored in that browser only, so it won't follow you to another browser or device.

What the setting doesn't cover

Feedback you give. If you rate a reply with thumbs up or thumbs down on a personal plan, OpenAI may use the entire conversation behind that rating for training, even after you've opted out. Skip the ratings if you want a chat kept out completely.
Codex environments. On a personal plan, the same setting covers your Codex tasks, but Codex has its own Include environments setting in Codex settings that the ChatGPT switch doesn't change.
Shared projects. Content in a project shared between personal accounts may be used for training only if the owner and every contributor have the setting turned on. One person opting out is enough to keep that project's data out.
GPT actions. When a custom GPT sends data to an outside service through an action, that service's own privacy policy applies to what it receives.
Memory is separate. Turning off training doesn't turn off memory, and vice versa. Our ChatGPT memory guide explains those controls.

For a single chat: use a temporary chat

If you're happy to help improve the models in general but want one particular conversation kept out, start a Temporary chat. While it stays temporary, it isn't used for training, doesn't show up in your history and doesn't create memories; OpenAI may still keep a copy for up to 30 days for safety reasons. If you later save it, it becomes a regular chat and your training setting applies to it. See our guide to using temporary chat in ChatGPT.

Good to know

Your history isn't touched. Turning training off doesn't delete or hide anything. Archiving a chat doesn't take it out of training either; only the setting does.
Support conversations follow the same switch. Chats with OpenAI Support may be used to improve OpenAI's services, including its models, while the setting is on.
Business feedback works differently. On Business, Enterprise and Edu, rating a response doesn't allow training on the conversation, and admins can switch response feedback on or off in the Admin Console.
OpenAI's privacy steps. When it does use eligible content, OpenAI says it first works to reduce the personal information in its training data.

If it doesn't work

The switch is missing or greyed out. In a managed workspace, the organization's settings apply, and for a linked teen account a parent or guardian may control it. Check with your admin or guardian.
The setting seems to have switched back on. If you were signed out when you changed it, it was saved only in that browser. Sign in and change it in your account settings.
You have more than one account. The setting belongs to each account. Change it in every account you use, or use the Privacy Portal request for each one.

Turning answers into work

If your chats are mostly plans you need to follow through on, Workbranch is a separate AI workspace with answers from Google Gemini, where chats sit in project folders with flags, due dates and a checklist for each answer. There's a 3-day free trial.

Sources

Data controls in ChatGPT (OpenAI Help Center, checked 18 October 2026)
How your data is used to improve model performance (OpenAI Help Center, checked 18 October 2026)
Temporary chat in ChatGPT (OpenAI Help Center, checked 18 October 2026)
Projects in ChatGPT (OpenAI Help Center, checked 18 October 2026)
Enterprise privacy at OpenAI (OpenAI, checked 18 October 2026)

Frequently asked questions

Will turning the setting off delete or hide my existing chats?

No. Your regular, archived and project chats all stay where they are. The setting only decides whether new conversations may be used for training. To remove content, delete the chats or your account separately.

Does the setting cover chats I had before I opted out?

OpenAI's pages describe the effect on new conversations after you opt out. They don't promise anything about content that was used before, so opt out as early as you can.

Can my parent or my employer control this setting?

Yes, in some cases. For a linked teen account, a parent or guardian can manage it through parental controls, and in a managed workspace the organization's settings apply instead of your own.