How to turn off ChatGPT training without deleting chat history
While signed in, open Profile → Settings → Data Controls and turn off “Improve the model for everyone.” New regular chats can remain in your history but will not be used to train ChatGPT. Temporary Chat is a separate no-history mode; neither option alone approves confidential data.
All contents
Turn off model improvement on your account
Check the account before changing the setting. If you use more than one account, check each separately.
Identify the personal workspace
Check the profile name or email. If work and personal accounts coexist, write down which one is being changed before opening Data Controls.
Turn off model improvement
On Web, use Profile → Settings → Data Controls → Improve the model for everyone. On mobile, open the sidebar, tap the profile icon and open Data Controls.
Confirm the synchronized state
Open the same account on the other device type and verify the setting. OpenAI says it is account-wide, so a mismatch usually means a different account is active.
Decide what information you can enter
Write down information that must not be entered and when to use an approved business workspace. Choose Temporary Chat only when its no-history behavior fits the task.
Choose the control for the outcome you need
| Goal | Current route | What still needs review |
|---|---|---|
| Keep history; stop training use | Settings → Data Controls → turn off Improve the model for everyone | Other processing and whether the input is permitted |
| Keep a chat out of history | Start a Temporary Chat | 30-day deletion description and abuse monitoring |
| Use approved organization data | Business service and workspace controls | Contract, access, retention and internal policy |
Turn off training use on Web or mobile
When signed in on the Web, click your profile icon, choose Settings, open Data Controls, and turn off “Improve the model for everyone.” OpenAI's current FAQ also gives a signed-out route through the ? icon in the lower-right corner and Settings.
On mobile, open the sidebar, tap your profile icon, choose Data Controls, and turn off the same setting. Record which account you changed; switching a browser profile or email address can put you in a different personal workspace.
One account setting covers Web and mobile
OpenAI says the training choice applies to the entire account and syncs across Web and mobile. You do not need a separate decision for each device when they use the same account.
You do need to repeat the check for every separate personal or work account. Confirm the profile name or email before pasting any material, especially when two browser profiles look similar.
History, training, memory and Temporary Chat are different
Turning off model-improvement use does not remove normal chats from history. Temporary Chats, by contrast, are not saved to history, do not create memories and are not used to train models.
As of the source check date, OpenAI says Temporary Chats are deleted from its systems after 30 days and may be reviewed for abuse monitoring. Use this mode when you do not want the conversation in your history; check permission separately before entering work information.
Personal Plus does not become Business
OpenAI says inputs and outputs from ChatGPT Business, Enterprise, Edu, Healthcare, Teachers and the API are not used for model training by default. Those services also have organization-specific contracts and controls.
Paying for a personal Plus plan does not grant the same contract or administration. For work data, record the product, workspace, retention controls, connected services and internal approval separately from the subscription decision.
Examples: regular chat or Temporary Chat
This example separates history, model improvement and data approval instead of treating them as one privacy switch.
- Draft you will revisit
- Regular chat with model improvement turned off; keep the thread in history
- One-off, non-sensitive question
- Temporary Chat if you do not want the conversation in history
- Client contract text
- Do not paste it into either mode without contract and organization approval
- Second personal account
- Repeat the setting check; synchronization does not cross accounts
The setting answers whether new personal-workspace chats can improve models. It does not answer whether the information is permitted, who can access it or which contract governs it.
Things to watch for
Changing the wrong account
Similar browser profiles can hide a different signed-in email. Verify the account before changing the setting or starting a sensitive workflow.
Treating history, memory and training as one control
Regular history can remain after training use is disabled. Temporary Chat changes history and memory behavior as well as model-improvement use.
Assuming Plus has Business data terms
A personal paid subscription does not replace an organization contract, administrator controls, retention review or internal approval.
Note the account and date you changed
Keep a short note of the account, setting date, when to use Temporary Chat and information you must not enter. Recheck after adding an account, changing the workspace type or seeing an official policy update.
Frequently asked questions
Will turning off ChatGPT training delete my chat history?
No. OpenAI says regular conversations can remain in history after you turn off Improve the model for everyone. Temporary Chat is the separate option for a conversation that should not appear in history.
Do I need to turn it off separately on my phone and computer?
OpenAI says the choice applies to the entire account and syncs across Web and mobile. Check every separate account, because a work email and a personal email do not share the setting.
Can I paste confidential data after turning training off?
Not on that basis alone. You still need to confirm the governing contract, retention and access conditions, connected services, confidentiality duties and organization policy. Use redacted or synthetic data when possible.
Scope and key points
Who this helps
- You want to keep regular ChatGPT history while stopping new personal-workspace chats from being used for model improvement
- You need to distinguish the training toggle from Temporary Chat and organization-level data controls
Limits of this guide
- You need approval to upload confidential, regulated or client-owned information—the toggle cannot provide that approval
- Your organization controls the workspace; check its contract, administrator settings and internal policy first
Key points
- The model-improvement control can be changed separately from chat history
- Temporary Chat has a different history and retention purpose
- The setting does not replace contract or organizational approval
What the sources say
The Improve the model for everyone setting is available under Data Controls and can be turned off while regular history remains
The setting applies across Web and mobile for the same account, while Temporary Chat has separate history, memory and deletion behavior
OpenAI states that listed business services and the API do not use organization inputs and outputs for training by default
References
- 1.Data Controls FAQ
OpenAI · Retrieved Aug 26, 2026
- 2.What if I want to keep my history on but disable model training?
OpenAI · Retrieved Aug 26, 2026
- 3.Business data privacy, security, and compliance
OpenAI · Retrieved Aug 26, 2026
About this article
On August 26, 2026, we rechecked OpenAI's Data Controls FAQ, personal-workspace training guidance and business-data policy, then separated the effect of each control and account type.
How we use AI in our research and writing ↗