ChatGPT Privacy Center launches; controls stay in Settings

Quick answer

OpenAI's September 21 announcement adds a Privacy Center that explains ChatGPT's privacy options and points to existing controls. It is a starting point for reviewing memory, data use, connected apps, and security—not a replacement for Settings. This article checks the documentation; it does not claim a hands-on rollout test or rank apps by privacy.

Download Chat AI Opens the official App Store or Google Play for your device.

A guide to the controls, not an automatic privacy change

Opening Privacy Center does not alter settings, erase data, or override workplace rules. OpenAI groups its guidance under Personalized chats & ads, Chat privacy, and Data use & access. A Manage action, when shown, opens the relevant existing control.

Sources: OpenAI Help Center, OpenAI Help Center

Where to find Privacy Center

OpenAI documents these entry points. If the new hub is missing, existing controls remain accessible through Settings.

Web: account menu → Help → Privacy center.
iOS or Android: Settings → Privacy Center or Privacy center.
Desktop: a Privacy center entry in Help, if present, opens it in a browser.

Sources: OpenAI Help Center

Three questions that need different controls

Start with the outcome you want. Keeping a conversation available, using it for personalization, and using it for model training are different decisions.

Training: Settings → Data controls → Improve the model for everyone governs whether new conversations help train OpenAI models. Turning it off leaves chats in history. If you submit response feedback, the associated conversation may still be used for training.
Memory: Settings → Personalization → Memory manages remembered context. Deleting a chat does not necessarily delete a separate saved memory. Removing information can require reviewing its other sources, including files and connected apps; telling the assistant not to mention it is not source deletion.
Temporary conversations: these are not used to improve models while temporary, but a copy may remain for up to 30 days for safety. Personalization is a separate starting choice. Saving the conversation makes it a regular chat subject to account settings. Third-party actions follow the recipient's policy.

Sources: OpenAI Help Center, OpenAI Help Center, OpenAI Help Center

An editorial exercise: map the requirement before uploading

Imagine a fictional workshop organizer who wants to revisit a draft next week but does not want new conversations used for model training. Which control addresses that requirement? Then add a second requirement: remove a venue address already saved in memory.

Sources: OpenAI Help Center, OpenAI Help Center

Apply the same questions when comparing Chat AI

A model name is not a privacy policy. For Chat AI, review its own privacy overview and the linked platform-specific policies before choosing a workflow. The overview explains that prompts and relevant context are processed by AppZone services and selected AI providers. This ChatGPT announcement does not establish equivalent controls in Chat AI.

Sources: Chat AI

Frequently asked questions

What readers usually ask

Do I need a paid plan to see Privacy Center?

No. Free is included in the documented rollout, although the hub may not have reached your account or app version.

Does turning off model improvement delete my saved chats?

No. Training preferences and deletion are separate. Regular chats remain in history unless removed through the applicable deletion controls.

Is Privacy Center the same as OpenAI's Privacy Portal?

No. Privacy Center explains options inside ChatGPT; the separate Privacy Portal website handles privacy requests, including requests for access to or deletion of personal data.

Evidence