How to Lock a ChatGPT Chat: Locked Chats Explained, the PIN and Face ID Rules, What Memory Does and Doesn't Do, What Happens If You Forget the PIN, and What Locking Doesn't Protect

What locked chats are, in OpenAI's words

The feature arrived with a post from @ChatGPT at 10:32 am PT on Tuesday, October 20 (342,000 views and 4,200 likes by Wednesday morning), a release-notes entry titled 'Introducing locked chats,' and the help article. There is no openai.com blog post and no executive commentary. The post: 'Move chats out of your regular history and search into a protected space you can unlock with a PIN on web and desktop, or a PIN, Face ID, Touch ID, or device unlock on supported mobile devices. New messages sent in locked chats won't update the memory used in your regular chats.' The release note adds the two lines that matter for privacy: 'Open them with your 4-digit account PIN or supported device authentication. Your retention and model-improvement settings stay unchanged.' Android Authority had spotted the feature in the app's code on August 21, two months before launch.

How to lock a chat, step by step

• 1. Open the chat's more-options menu (•••) in the sidebar or at the top of the conversation.
• 2. Select Lock chat, then Lock chat again to confirm.
• 3. First time only: create a PIN. Choose Create a PIN, enter four digits, enter them again. The PIN belongs to your OpenAI account, not to a workspace or device, so it follows you across web, desktop, and phone.
• 4. On a phone, optionally turn on Face ID, Touch ID, or your device unlock. OpenAI: 'Device authentication is optional and applies only to the device where you enable it.' You still need the PIN to change the PIN.
• 5. Find it later under Projects > Locked chats. The sidebar may show the Locked chats project while the titles inside stay hidden. One lag to know about: other devices with ChatGPT already open 'may continue to show the chat title until their conversation lists refresh.'

Relocking. On web and desktop, locked chats relock when you've been away from that tab or window, and web access also ends if you refresh or close the tab. On iOS and Android, access ends after one minute away from the locked chat or the Locked chats area, or one minute in the background. A Lock now button ends it immediately. Too many wrong PINs temporarily disables access; OpenAI doesn't publish the count or the cooldown. To move a chat back, open it, choose ••• then Unlock chat (Unlock on iOS), and confirm; it returns to regular history, and if it came from a project it does not go back to that project. Memories created while it was locked stay in the locked pool rather than joining regular memory.

What memory does and doesn't do inside a locked chat

This is where the launch wording and the aggregator posts blur things. 'Separate memory' is half right. From the help page, exactly: 'Locked chats can use relevant information from regular Memory.' So ChatGPT still knows your job, your kids' names, and your preferences inside a locked chat. What changes is the direction of flow: 'Information from new messages sent after you lock a chat can be remembered for use in other locked chats,' and 'Information remembered only within locked chats does not carry into regular chats.' Locking also doesn't erase anything ChatGPT already remembered from the conversation before you locked it. If your workspace has Memory disabled, locked chats can't use or create memories at all. Practical translation: locking stops a sensitive topic from surfacing in tomorrow's unrelated answer, which is the thing most people want; it does not make the chat anonymous to the model, and it does not undo what was said before the lock.

What locking does not protect

• It isn't encryption. The help page says locking does not 'create end-to-end encryption or a separate encrypted vault.' OpenAI can see locked chats exactly as it sees any chat. Anyone who can open one 'can still copy its contents.'
• Training is unchanged. 'Locked chats follow your existing model-improvement setting.' If 'Improve the model for everyone' is on, locked chats are still eligible. Turn it off separately; our September guide has the steps.
• Retention is unchanged. Locking does not 'change how long OpenAI retains the chat.' Deleting one later follows the standard policy, permanent deletion within about 30 days barring legal or security holds. For the record, OpenAI's NYT litigation page says the indefinite-retention order ended September 26, 2025, and has not been updated since.
• Your employer can still see it. Workspace admins can't disable locked chats, unlock yours, or reset your PIN. But in Enterprise and Edu, locked chats 'are still included in conversation message logs available through the OpenAI Compliance Platform,' and locking does not 'override applicable workspace policies, safety requirements, or legal and compliance obligations.' On Business, workspace policies and retention also stay as they were.
• Shared links survive. If you shared a chat before locking it, 'locking it does not delete the link.' Delete the link yourself.
• Ads can follow you in. 'If ads are available for your account, they may also appear in locked chats.' The $8 Go plan 'may include ads.'
• Small print. On Android, screenshots of locked chats 'may appear blank or black.' Voice works in locked chats on web and mobile but not in the desktop app. Files you lock stay hidden from Space and its search, but files from connected sources keep their existing access rules.

If you forget the PIN, everything locked is deleted

Read this before you lock anything you can't lose. Forgetting the PIN doesn't delete anything by itself. Resetting it does: 'Resetting your account PIN permanently deletes all locked chats, remembered information stored only within locked chats, and files or other content protected only by your previous PIN. This cannot be undone, and OpenAI Support cannot recover the deleted content.' The path is Settings > Security and login > Account PIN > Reset, verify your identity, then a button labeled Reset PIN & Delete (web, desktop, Android) or Delete all and reset (iOS). There is no 'send me a reminder' option, and biometrics don't help, because changing or resetting the PIN requires the PIN. Write the four digits down somewhere you'd actually find them.

What you can't lock, and why Lock chat isn't showing

Can't be locked: temporary chats (while temporary), archived chats, group chats, chats from a shared link, chats with scheduled tasks, meeting recordings, or background tasks, chats whose responses are saved as project sources, chats in the earlier Health experience, and all Codex and local Work conversations. Locking a chat that lives in a project pulls it out of the project, so it stops using that project's instructions, files, and memory. Once locked, a chat can't be archived, pinned, moved, or shared, and ChatGPT can't create or update a scheduled task from it. Not available at all on ChatGPT for Healthcare. Not showing yet: the help page says the feature 'is rolling out gradually. If you don't see Lock chat in an eligible chat's more options menu (•••), the feature may not be available for your account yet.' There is no setting to turn it on, no region list, and no stated difference between iOS and Android timing. Update the app, check the ••• menu on a plain, unarchived chat, and wait; the toggle appears when your account gets it.

Locked chats vs Temporary Chat vs Claude, Gemini, Grok, and Meta

Locked chats are for conversations you want to keep but hide. Temporary Chat is for conversations you don't want kept at all. They are not interchangeable, and a temporary chat can't be locked while it's temporary. Here's how the field compares as of October 21, from each vendor's help pages.

Two observations. OpenAI is the only one of the five offering a lock on saved chats; everyone else offers only the don't-save mode, which ChatGPT also has. And the name has history: WhatsApp's Chat Lock, which puts conversations in a 'Locked chats' folder behind your device passcode or biometrics, launched May 15, 2023; Google Photos has had a Locked Folder for years. OpenAI borrowed a pattern people already understand, which is probably why the post traveled.

Our read

This is a good, overdue feature with a launch post that undersells its limits and a help page that, to OpenAI's credit, oversells nothing. The thing people actually want, 'don't let my health scare or my job search leak into my regular answers and don't show it on the shared laptop,' is exactly what it does. What it doesn't do is make anything private from OpenAI, from training if you've left that on, from a legal hold, or from your employer's compliance export, and the words 'protected space' in the announcement will lead some people to assume otherwise. The PIN-reset-deletes-everything rule is the sharpest edge in the product and should have been in the post. For ChatGPT users the setup we'd suggest is: turn off model improvement in Data controls once, use Temporary Chat for things you never want saved, and use locked chats for the things you do. For the rest of the chatbot field, this is now a gap: Claude's incognito chats and Gemini's temporary chats are good don't-save modes, and neither has an answer for the saved-but-hidden case. Related: Can Claude and ChatGPT report you to the police? and Did ChatGPT leak my photos?