New icon in the account menu, bold headline: "Privacy Center." Sounds like a big step for privacy. It isn't — not in the sense the name promises, anyway.
A few days ago, OpenAI rolled out a new Privacy Center for ChatGPT. It bundles all privacy-related info in one place: memory, temporary chats, personalized ads, model training, connected apps, account export and deletion. Sounds useful. It is. Except: there are no new settings in it. Not a single new switch. The Privacy Center just explains the switches that already existed, and links to them.
What's actually in it
Three categories, each linking to the real setting:
- Personalized chats & ads — how memory works, plus the genuinely useful note that deleting a chat does NOT automatically delete memories saved from it. If you've ever wondered why ChatGPT still "remembers" something after you cleared your history — this is why.
- Chat privacy — temporary chats, which don't show up in history and aren't used for training.
- Data use & access — the "Improve the model for everyone" toggle (does your chat history get used for training?), connected apps and plugins, data export and deletion.
On the web, find it via the account menu → Help → Privacy Center, or the matching address on chatgpt.com. In the app, it sits under Settings. Important for anyone without a subscription: the Privacy Center is available for Free, Go, Plus, Pro, and Business alike — no extra cost, no subscription required. Only ChatGPT Enterprise, Edu, and the healthcare variant are left out, since admins manage those settings centrally anyway.
The catch
"Privacy Center" sounds like a control room. It's really more of a table of contents. OpenAI says so itself: the actual settings stay exactly where they were — in regular ChatGPT settings. If you already knew where the training toggle was hiding, you get nothing new here. If you didn't, you'll now find it explained in one place instead of buried across three nested submenus.
That's not a scandal. Just a mismatch between what the name promises and what the feature delivers.
Keeping it in proportion
Google and Meta have had privacy overviews like this for years. OpenAI catching up is overdue, but not a reason for outrage in either direction. A dashboard doesn't automatically make a company more privacy-friendly — it just makes visible what was already there. For a broader sense of how much (or little) Big Tech knows about you overall, our reality-check comparing privacy across the major AI providers is a good next read.
What you can actually do
Two minutes, worth spending:
- Open the Privacy Center (web: account menu → Help → Privacy Center; app: Settings → Privacy Center).
- Check the "Improve the model for everyone" toggle — turn it off here if you don't want your chats used for training.
- Look through memory to see what ChatGPT has stored about you. You can delete individual memories without wiping your entire chat history.
- Use temporary chats for anything you don't want remembered — the default chat retains things, the temporary one doesn't.
None of this is new. But now you can actually find it without having to go looking.