OpenAI has introduced a new ChatGPT feature that allows users to lock individual conversations behind a four-digit PIN or biometric authentication, adding a layer of privacy for people who share devices or want to prevent others from casually accessing sensitive discussions.
Announced on Tuesday, the feature, called Locked Chats, allows users to separate selected conversations from their regular chat history and protect access using Face ID, Touch ID or a PIN on supported devices.
The rollout comes as ChatGPT is increasingly used for personal conversations alongside work, research and everyday tasks. Users may discuss private matters, share sensitive information or plan surprises in chats they do not want others to open when borrowing their phones or computers.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
OpenAI said the feature is designed for situations where users want to hand their device to someone else without exposing a particular conversation. Unlike Temporary Chats, which provide a separate session that does not appear in regular chat history, Locked Chats remain accessible to the account holder after authentication.
The distinction gives users a way to retain conversations while restricting casual access to their contents. However, the feature does not create a separate encrypted vault or fundamentally change how ChatGPT stores and processes conversation data.
How Locked Chats Work
Once a conversation is locked, it moves into a dedicated Locked Chats section, separate from regular chat history, pinned conversations, and chat search results. Users can return to those conversations by unlocking the protected section with their PIN or supported biometric authentication.
The feature is intended to provide an additional barrier against someone opening a private conversation while using the same device. It also separates locked conversations from the normal browsing experience within ChatGPT, making them less visible to someone navigating the user’s chat history.
OpenAI has also introduced a separate memory arrangement for Locked Chats when Memory is enabled. Conversations within the locked section can share their own memory, allowing information from those chats to inform other conversations in the same section.
New messages sent after a chat is locked can update that separate memory but do not update the user’s regular Memory. This is intended to prevent information introduced in locked conversations from carrying over into ordinary chats.
The separation is necessary for users who want to keep certain topics isolated from their broader ChatGPT interactions. Nevertheless, it should not be confused with complete data isolation or encryption. The feature controls access to conversations within the app rather than establishing an entirely separate security environment.
OpenAI also warns that resetting a PIN permanently deletes all locked chats, files, and other content protected by the previous PIN. Users should therefore understand the consequences before initiating a reset, particularly if they rely on locked conversations to retain important information.
A Privacy Layer, Not End-to-End Encryption
Despite the additional access controls, OpenAI has outlined several limitations that distinguish Locked Chats from stronger forms of data protection.
Locking a conversation does not introduce end-to-end encryption or create a separate encrypted vault. It also does not change ChatGPT’s data-retention practices, remove information the system remembered before a conversation was locked, or modify the user’s existing “Improve the model for everyone” setting.
These limitations matter because restricting access to a conversation on a device addresses only one part of the broader privacy question.
A PIN or biometric check can help prevent someone who temporarily has access to a device from opening a protected chat. It does not, by itself, determine how the service retains conversation data, how existing memory is handled or whether the user’s data settings permit conversations to be used to improve models.
Users who want to protect sensitive information should therefore distinguish between access control and data handling. Locking a chat may reduce the risk of casual exposure to friends, family members, or other people using the same device, but it does not replace reviewing account privacy settings or understanding the service’s data policies.
The treatment of information remembered before a chat is locked is relevant. Because locking a conversation does not erase existing Memory, information already retained outside the locked section may remain available to regular conversations, depending on the user’s settings and what ChatGPT has remembered.
The separate memory arrangement applies to new information from locked conversations, according to OpenAI. Users should not assume that locking a chat retroactively removes previously remembered details.
Some ChatGPT Conversations Cannot Be Locked
OpenAI has excluded several conversation types from the new feature. Locked Chats is not available for Temporary Chats, archived conversations, group chats, conversations created through shared links, chats with scheduled tasks, or chats containing meeting recordings.
The exclusions mean that users cannot apply the same protection uniformly across every type of ChatGPT interaction. Anyone seeking to restrict access to a particular conversation will need to check whether that conversation is eligible for locking.
The feature is being rolled out progressively, beginning on Tuesday, so it may not appear immediately for every user.
OpenAI says Locked Chats will be available on ChatGPT Free, Go, Plus, Pro, Business, Enterprise, and Edu plans. The feature is not available on ChatGPT for Healthcare.
What the Feature Means for ChatGPT Users
The introduction of Locked Chats addresses a practical privacy concern as conversational AI becomes part of users’ personal and professional routines. ChatGPT conversations can contain information people prefer not to display when sharing a device, even when they have no intention of deleting the underlying discussion.
Previously, users seeking to keep a conversation out of view had to consider alternatives such as Temporary Chats, which do not appear in regular chat history, or avoid leaving sensitive discussions accessible in their usual conversation list. Locked Chats offers another option by preserving access while adding an authentication barrier.
Its value, however, depends on what users expect privacy to mean. For preventing someone from casually opening a conversation on a shared device, the feature provides a useful additional control. For concerns about data retention, model improvement, or the broader handling of sensitive information, users will still need to consider the relevant account settings and OpenAI’s data policies.
However, locked Chats give users more control over who can open selected conversations within ChatGPT. But it does not eliminate the wider privacy considerations associated with storing personal information in an online AI service. The feature is best understood as an additional access-control measure, rather than a guarantee that locked conversations are isolated from every other aspect of the platform’s data systems.