chatgptsecuritytutorialshow-to

How to Use ChatGPT Privacy Center on Web and Mobile

Open ChatGPT Privacy Center, review memory, ads, temporary chat, and model training, then jump to the setting that actually changes each control.

September 22, 2026 Yum News

ChatGPT now has a Privacy Center inside the product. It does not replace Settings. It explains what each privacy control does, then sends you to the existing toggle when you want to change it.

OpenAI started the rollout on September 21, 2026 for signed-in Free, Go, Plus, Pro, and Business users on the web and on the iOS and Android apps. Enterprise, Edu, and ChatGPT for Healthcare are not in this rollout. If the menu item is missing, your plan, region, or app version may still be waiting.

Opening Privacy Center does not change settings, delete chats, or override a workspace policy. Treat it as a map, not a reset button.

Laptop and phone on a desk used for account security settings

What Privacy Center is — and is not

Privacy Center lives inside ChatGPT. The older Privacy Portal is a separate website for formal requests such as a data download or a request to remove personal data from model responses.

Your real controls stay under Settings. Privacy Center groups those controls into three buckets:

Personalized chats and ads
Chat privacy
Data use and access

Select a topic to read the short explainer. If Manage appears, that button opens the live setting. Some topics offer a different action, such as New temporary chat.

If you also draft in Word, remember that the ChatGPT for Word sidebar does not share history with chatgpt.com. Privacy Center on the main app does not rewrite add-in behavior.

How to open Privacy Center

Web

Sign in at ChatGPT.
Open your account menu.
Select Help.
Select Privacy center.

iOS

Open the ChatGPT app.
Open Settings.
Select Privacy Center.

Android

Open the ChatGPT app.
Open Settings.
Select Privacy center.

Desktop app

If the Help menu lists Privacy center, select it. OpenAI opens the center in your browser.

If you do not see the item, update the app and check that you are signed in on a supported plan.

Personalized chats and ads

This category covers Memory, personalized ads, and location.

Memory. ChatGPT can save facts from chats and reuse them later. Deleting a chat does not automatically delete a memory created from that chat. Use Manage to open Memory settings if you want to review or clear saved items.

Personalized ads. Ads can appear for eligible Free and Go users. Business, Enterprise, Edu, and ChatGPT for Healthcare do not show ads. Review ad controls only if your plan is eligible.

Location. ChatGPT can use location for relevant search results when sharing is available. OpenAI says it does not separately store your precise location outside the information already in a conversation and its context. Precise location sharing is not available in ChatGPT Enterprise workspaces.

Person reviewing security options on a laptop in a quiet office

Chat privacy and temporary chats

Select Chat privacy, then Temporary chat.

Temporary chats stay out of your history. They do not create or update memories. While a chat remains temporary, OpenAI does not use it to improve models.

You can still choose whether a temporary chat is personalized before it starts. A personalized temporary chat can use existing memories, custom instructions, and plugins. Turn personalization off at the start if you want a clean slate.

If you save a temporary chat, it becomes a regular chat. From that point it follows your account history, personalization, and model-improvement settings.

To start one from Privacy Center: open Temporary chat, then New temporary chat.

Data use, training, export, and account security

Plugins and apps. This topic explains what connected apps can read and which permissions they need. A workspace admin can limit which apps appear.

Multi-factor authentication. Use this path to add a second sign-in step. Follow OpenAI’s MFA help article for the exact setup screens.

Model improvement. On eligible personal accounts, Improve the model for everyone decides whether new conversations help train OpenAI models. Turning it off does not delete chats or remove them from history.

On web:

Account menu → Settings → Data controls.
Turn off Improve the model for everyone.
Select Done.

On iOS and Android:

Open the sidebar and tap your profile.
Open Settings → Data controls.
Turn off Improve the model for everyone.

You can also select Do not train on my content in the Privacy Portal. OpenAI says either path is enough for ChatGPT conversations and Codex tasks on a personal plan. You do not need both.

By default, OpenAI does not use content from ChatGPT Business, Enterprise, Edu, or ChatGPT for Healthcare workspaces to train models.

Export or delete. Signed-in Free, Go, Plus, and Pro users can request an export from Settings → Data controls, or through the Privacy Portal. Self-service export is not available in Business, Enterprise, or Healthcare workspaces. Some Edu workspaces allow export only if an admin turns it on. Self-service account deletion is not available in managed workspaces.

Voice and video sharing can have separate controls. Check the ChatGPT Voice help pages if you use those features.

Business workspaces

If you use ChatGPT Business, your organization can limit which privacy features you see and how workspace data is retained or reviewed.

Privacy Center does not override admin controls, retention policies, or authorized access to workspace data. If a Manage button is missing, that is often policy, not a bug.

A 10-minute checklist

Open Privacy Center and confirm your plan is in the rollout.
Read Memory, then open Memory settings if you keep personal facts in chats.
Check ads only if you are on Free or Go.
Start one temporary chat for a sensitive draft, then close it without saving.
Turn off Improve the model for everyone if you do not want new personal chats used for training.
Confirm MFA is on.
Request an export only if you need an archive. Store the download like any other account dump.

Do not paste account numbers, government IDs, or medical record numbers into a regular chat when a redacted snippet will do. OpenAI Academy’s short privacy clip above walks through that habit with a utility bill example.

Conclusion

Privacy Center is a guided index of controls that already existed in Settings. Use it to find Memory, ads, temporary chat, model training, MFA, and export options without hunting through menus.

Change the setting you care about, then leave the center. Opening it does nothing by itself. For formal access or deletion requests that Settings cannot finish, use the Privacy Portal instead.