How to Lock Your ChatGPT Chats With a PIN or Face ID (2026)

OpenAI has given ChatGPT something it never had before: a way to hide selected conversations behind a lock. Announced on 20 October 2026, Locked Chats lets you seal any chat with a 4-digit PIN — or with Face ID, Touch ID or your device unlock on phones and tablets — so it disappears from your history, pinned chats and search until you authenticate. This guide walks through how to lock a chat, how the PIN and biometrics work, what the lock does not protect you from, and the one trap that can permanently delete everything you locked.

Quick answer: OpenAI's Locked Chats (20 Oct 2026) hides any chat behind a 4-digit PIN or Face ID. How to lock a conversation, enable biometrics, and avoid the PIN-reset trap.

What are Locked Chats in ChatGPT?

Locked Chats is a privacy feature for shared screens and shared devices: the person who borrows your phone to “look at one chat” can no longer scroll into everything else. When you lock a conversation, it moves out of your regular chat history, pinned conversations and search results into a dedicated “Locked chats” section under Projects. It stays saved in your account — you can come back to it any time — but opening it requires your 4-digit account PIN, or a biometric check on a supported mobile device.

There is also a memory split worth knowing. With Memory turned on, a locked chat keeps its own memory shared only across other chats inside the locked section; new messages sent after locking update that separate memory rather than your regular Memory, so nothing new from a locked chat carries over into your unlocked conversations. That separation only applies from the moment you lock it — nothing ChatGPT already remembered before the lock gets erased.

How to lock a ChatGPT chat: step by step

The feature is rolling out gradually, so if the option is not visible yet, update the app and check again. On an account that has it:

  1. Open the conversation you want to lock in ChatGPT (web, desktop or mobile).
  2. Open its options menu — the three dots next to the chat title.
  3. Select “Lock chat.” ChatGPT will ask you to set up or enter your 4-digit account PIN.
  4. Confirm the PIN. The chat leaves your regular history immediately and moves into the Projects > Locked chats section.
  5. On iPhone or Android, enable Face ID, Touch ID or your device’s unlock method when prompted, so future access is one biometric check instead of a PIN.

To unlock, open the Locked chats section and authenticate with your PIN or biometrics. That is the whole mechanism — there is no master-password layer beyond the account PIN itself.

How to unlock locked chats with Face ID, Touch ID or PIN

On the web and desktop apps, the lock is PIN-only: you type the 4-digit code each time you open the section. On phones and tablets, OpenAI goes a step further — alongside your device passcode, you can unlock with Face ID on iPhone or a fingerprint sensor on Android, plus whatever unlock method your device supports. On Android there is one extra wrinkle: screenshots of locked chats can appear blank or black, because the app hides protected content from the screenshot layer. If you need to keep a record of something in a locked chat, copy the text out before you lock it.

What Locked Chats will not do

OpenAI has been unusually explicit about the limits, and this is where the honest reading matters. Locking a chat:

In short: Locked Chats stops a person looking over your shoulder, not a party with access to your account or to OpenAI’s data. Treat it as a screen-privacy feature, not a secrecy feature.

The PIN-reset trap: forgetting your PIN deletes everything

This is the single most important thing to know before you lock anything. If you forget your account PIN and reset it, every locked chat, file and other item tied to the old PIN is permanently deleted. OpenAI warns there is no undo, and OpenAI Support cannot recover the deleted content. The lock is deliberately destructive on reset — there is no “forgot PIN, email me a link” recovery path.

Practical rules that follow from that design: never store your only copy of anything important inside Locked chats; if you must lock something valuable, keep a copy elsewhere first; and treat the PIN like a password — write it down somewhere safe, because a reset is not a mild inconvenience, it is total loss of the locked content.

Locked Chats vs Temporary Chats vs Archived Chats

ChatGPT now has three different ways to make a conversation less visible, and they do very different things:

Locked Chats Temporary Chats Archived Chats
Saved in your account? Yes — saved and locked No — never saved to history Yes — saved, moved out of the sidebar
Hidden from history/search? Yes, until PIN/biometric unlock Yes — nothing is saved at all Hidden from the sidebar, still in account search
Unlock mechanism 4-digit PIN / Face ID / Touch ID N/A Unarchive, no authentication
Memory behaviour Separate locked-section memory No memory updates at all Normal memory
Can it be locked? — No, not eligible No, archived chats cannot be locked
Biggest risk PIN reset deletes everything permanently Nothing is recoverable by design None — it is just filed away

Also note what cannot be locked at all: temporary chats, archived chats, group chats, chats opened from a shared link, chats with scheduled tasks, chats containing meeting recordings, Codex conversations — and the feature is not offered on ChatGPT for Healthcare.

When locking is worth it — and when it is not

Lock a chat when you share a device or screen regularly — family phones, work laptops on a projector, a demo where you hand the phone to someone. Planning a surprise, discussing something personal, or keeping client names out of casual view are the exact cases OpenAI designed this for.

Do not rely on it when the threat is anything more serious than a curious glance. If you need a conversation excluded from training, turn off the training setting separately — locking does nothing there. If you need it gone from OpenAI’s systems, delete it and check your retention settings — locking keeps it saved. And if the PIN is forgettable, the lock is riskier than no lock, because a reset wipes the content with no recovery.

Pair this with the rest of ChatGPT’s 2026 privacy surface: if you use the ChatGPT browser agent in Chrome, remember that Work mode can fill forms and send messages on your behalf — a locked chat does not restrict what the agent can do elsewhere. And the copy-paste Chrome prompts we tested are safe to run in either mode, since the lock only gates visibility, not capability.

Frequently asked questions

Is Locked Chats available on the free ChatGPT plan?

Yes. OpenAI is rolling it out across Free, Go, Plus, Pro, Business, Enterprise and Edu plans. It is not available on ChatGPT for Healthcare. The rollout started 20 October 2026 and is gradual, so the option may not appear in your account immediately — update the app and check again.

Why can I not see the “Lock chat” option?

The most likely reason is that the gradual rollout has not reached your account yet. Update to the latest version of the app (or reload the web app) and look again. It also will not appear on ineligible chats: temporary chats, archived chats, group chats, shared-link chats, scheduled-task chats, meeting recordings, and Codex conversations.

What happens if I forget my Locked Chats PIN?

Resetting the PIN permanently deletes every locked chat, file and other content tied to the old PIN. OpenAI says this cannot be undone and Support cannot recover it. Store the PIN somewhere safe before you lock anything important.

Does locking a chat stop ChatGPT from training on it?

No. Locking does not change the “Improve the model for everyone” setting, does not alter data retention, and does not erase anything already in Memory. To control training, change the setting separately under your data controls.

Is there end-to-end encryption on locked chats?

No. OpenAI says explicitly that locking is an access barrier in the app interface, not end-to-end encryption and not a separate encrypted vault. It protects against someone physically handling your unlocked device, nothing more.

Sources and methodology

Feature details verified 23 October 2026 against OpenAI’s 20 October 2026 announcement as reported by Digital Trends, The Hindu (21 Oct 2026), Pondero (summarising TechCrunch’s original 20 Oct report) and AI Weekly. Claims rest on those reports’ reading of OpenAI’s announcement and support documentation. Method: desk research — this feature was not hands-on tested (the rollout is gradual and the option may not be visible in every account), so interface steps follow the documented flow. Our testing standards are at how we test AI tools.

Leave a Reply

Your email address will not be published. Required fields are marked *