ChatGPT Gets a Privacy Lock: PIN and Biometrics Seal Off Your Most Sensitive Chats

ChatGPT Gets a Privacy Lock: PIN and Biometrics Seal Off Your Most Sensitive Chats

Any single chat can now be locked behind a PIN or your face

OpenAI on October 20, 2026 announced Locked Chats for ChatGPT, letting users seal off individual conversations from anyone who picks up their phone or borrows their computer. On the web and desktop, the lock is a user-set four-digit PIN; on mobile, supported devices can use Face ID, Touch ID, or Android fingerprint sensors. According to DigitBin's walkthrough of OpenAI's help center, locking a chat for the first time requires creating a four-digit PIN and entering it twice. The PIN belongs to the OpenAI account rather than to one device, so it works on every supported device where the user is signed in.

Locked chats vanish from history, pins, and search into their own vault

A locked conversation disappears from regular chat history, pinned chats, and conversation search, and moves into a dedicated "Locked chats" section under Projects in the sidebar. It can only be reopened after passing PIN or biometric authentication. Unlike Temporary Chats, which run in a separate session that never appears in history, a locked chat stays with the account and can be revisited at any time. The idea is simple: the conversation is preserved but invisible until its owner unlocks it.

Even ChatGPT's memory is split so locked chats cannot leak sideways

To stop private details from leaking through the assistant's memory, OpenAI gave locked chats their own isolated memory. Citing the company, PCMag and Yahoo Tech report that with Memory enabled, the Locked Chats project keeps a separate memory shared across conversations inside that section. New messages sent in a locked chat can update that shared locked memory but never update regular account memory, so fresh details revealed in a locked thread do not surface later in ordinary chats.

Lose the PIN and hit reset, and everything locked is gone for good

The sharpest edge of the feature is its PIN-reset policy. Multiple outlets, including PCMag, DigitBin, and Cryptopolitan, cite OpenAI's warning: changing the PIN while knowing the current one keeps every locked chat intact, but resetting a forgotten PIN permanently deletes all locked chats, along with files and memories protected by the old PIN, and OpenAI Support cannot recover any of it. DigitBin lays out the reset path as Settings, then Security and login, then Account PIN, then Reset, confirmed with "Reset PIN & Delete" on web, desktop, and Android, or "Delete all and reset" on iOS. Users should memorize the code and keep anything irreplaceable outside locked chats.

No end-to-end encryption, and OpenAI says so up front

Lifehacker Australia and Firstpost, citing OpenAI's support page, stress what the lock is not: it adds no end-to-end encryption, creates no separate encrypted vault, does not change how long chats are retained, does not erase information ChatGPT memorized before the chat was locked, and does not alter the "Improve the model for everyone" data setting, which remains switchable under Settings and Data controls. In short, this is screen-level privacy against people nearby, not cloud-level encryption against data retention.

Sharers and borrowers gain most, while six chat types stay unlocked

The feature is aimed squarely at people who share phones or computers, where a family member, partner, or colleague glancing at the screen could otherwise stumble onto sensitive threads. Citing official documentation, the Spanish-language outlet iaenunminuto notes six categories of chats that cannot be locked: temporary chats, archived chats, group chats, chats created from a shared link, chats with scheduled tasks, and chats containing meeting recordings. The same outlet and others add that the rollout is gradual, so some accounts may not see the entry point yet.