OpenAI adds PIN and biometric locks for private ChatGPT conversations
The new Locked Chats feature allows users to hide specific conversations behind a four-digit PIN or Face ID to prevent casual browsing of personal history.
The short version
- Locked chats are moved to a dedicated area under Projects and are hidden from regular history and search results.
- Users can authenticate using a four-digit PIN or device-specific methods like Face ID, Touch ID, or fingerprint recognition.
- The feature is rolling out to Free, Plus, Pro, Business, Enterprise, and Edu users across web and mobile platforms.
- Locking a chat removes it from its original project, causing it to lose access to that project's specific instructions and files.
- Resetting a forgotten PIN permanently deletes all locked chats and their associated memories, which cannot be recovered by support.
Why it matters
This gives users a way to keep sensitive personal planning or private advice out of sight when sharing a device with others. It balances the convenience of saving a long-term conversation with the need for basic visual privacy.
How Locked Chats work
OpenAI has introduced a new feature called Locked Chats designed to keep specific conversations private from anyone glancing at your screen. When you lock a conversation, it is moved into a separate area located under the Projects menu. This means the chat will no longer appear in your standard history, your pinned chats, or your regular search results.
To activate this, users can select the three-dot menu on a conversation and choose to lock it. If it is the first time using the feature, the system will prompt you to create a four-digit PIN. This PIN belongs to your entire OpenAI account, allowing you to use the same code across different devices where you are signed in.
While the content remains accessible to you, it stays hidden until you manually open the locked section. You can continue an existing conversation or start a brand-new one directly within this protected area. Opening a locked chat only gives you temporary access; it does not permanently move the conversation back into your main history.
Biometrics and device access
For users who find typing a PIN tedious, OpenAI supports biometric authentication on compatible mobile devices. iPhone users can enable Face ID or Touch ID, while Samsung Galaxy and Google Pixel users can use their specific device unlock methods, such as fingerprint recognition. These settings are managed within the security and login section of the app settings.
It is important to note that device authentication is specific to the hardware where it is enabled. Turning on a fingerprint lock on a Pixel phone will not automatically enable that same method on a different phone. However, the four-digit account PIN remains available as a backup for all supported devices and the web interface.
The app is designed to automatically relock the conversation for security. On mobile devices, access ends automatically after you have been away from the locked area for one minute. On the web version, closing the tab or refreshing the page will also end the authorized session and relock the content.
Technical limitations and memory
There are several specific limitations to keep in mind regarding how these locks interact with other features. Locking a conversation removes it from any existing ChatGPT Projects. This means the chat loses access to the specific instructions, files, and memory associated with that project, even if you unlock it later. Additionally, the feature does not provide end-to-end encryption or create a separate encrypted vault.
The way the AI remembers information also changes slightly when a chat is locked. If the AI remembered something before the lock, that data stays in the regular memory. However, any new information the AI learns from messages sent within a locked chat stays only within the Locked Chats memory area. This means regular chats will not automatically receive those new memories.
Furthermore, locking a chat does not revoke any shared links created before the lock was applied. If you previously shared a conversation link, anyone with that link can still view the content. To stop access, you must manually delete the shared link. Files uploaded to a locked chat are protected, but files stored in other parts of your Space or connected services are not.
Rollout and account types
OpenAI is rolling out Locked Chats gradually to a wide range of user tiers. This includes users on the ChatGPT Free, Go, Plus, Pro, Business, Enterprise, and Edu plans. The feature is available on the web, the iPhone and Android apps, and the new desktop applications for both Mac and Windows computers.
Some types of conversations are ineligible to be locked by the system. This includes Temporary Chats, archived chats, group chats, and conversations created from shared links. Additionally, chats involving scheduled tasks, meeting recordings, or background tasks cannot be secured with a PIN. If you do not see the option to lock a chat, it may be because your account hasn't received the update yet or the specific chat type is excluded.
For workplace accounts, standard organizational retention and compliance policies still apply. For example, Enterprise and Edu conversation logs may still be accessible through official compliance tools. While workspace administrators cannot see your PIN or unlock your chats, they may still have access to the data through these administrative systems. Users should still avoid sharing highly sensitive data like passwords or Social Security numbers in any chat.
Managing your security
Users have the ability to manage their PINs without losing their data. If you know your current PIN but want to change it, you can do so through the security and login settings on the web or mobile app. This allows you to update your credentials while keeping all your locked conversations intact.
However, if you forget your PIN, the recovery process is strict. To reset a forgotten PIN, you must verify your identity and confirm the permanent deletion of all your locked chats. OpenAI states that this deletion cannot be undone, and their support team is unable to recover any content, memories, or files that were protected by the old PIN.
For users who want to prevent their data from being used for training, OpenAI offers a separate setting. You can turn off the 'Improve the model for everyone' option in your data controls. While Locked Chats hide your history from people looking at your screen, turning off model training ensures your new conversations are not used to train OpenAI's underlying AI models.
Comparison with Temporary Chat
It is helpful to distinguish Locked Chats from the existing Temporary Chat feature. Temporary Chats are designed for one-time discussions that you do not want to save at all. They do not appear in your history and do not create new memories, making them ideal for quick tasks that don't require a long-term record.
Locked Chats are better suited for ongoing personal projects, such as planning a complex event or working through a personal issue over several days. They allow you to maintain a continuous thread of conversation while keeping it hidden from the main sidebar. This provides a middle ground between a completely public history and a completely deleted one.
Ultimately, the company suggests treating these locks as an access-control convenience rather than a total privacy guarantee. While the PIN helps keep curious eyes away from your sidebar, your overall privacy still depends on the specific settings you choose and the type of information you decide to share with the AI. Users should still be mindful of the data they provide to the platform.
Sources and links
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by Fox News on MSN · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
