OpenAI has enabled a lock feature for ChatGPT conversations that requires a four-digit PIN or biometric authentication to access.
How the lock works
The new Locked Chats section sits separate from regular history, pinned conversations, and search results. Unlike Temporary Chats, which vanish from view after the session ends, users can return to locked conversations at any time.
OpenAI describes the tool as useful when handing a device to friends or family. A user might want to show a different chat while keeping a personal discussion or a surprise plan hidden from view.
While this adds a layer of privacy for the current session, it does not change how sensitive information is stored or handled by the wider system. The feature does not create end-to-end encryption, nor does it act as a separate encrypted vault.
Locking a chat does not alter how long OpenAI retains the data, remove information the model remembered before locking occurred, or change the global “Improve the model for everyone” setting.
Memory and data rules
When Memory is enabled, Locked Chats maintains its own separate memory shared only across conversations within that specific section. New messages sent after locking a chat can update this local memory but do not update regular Memory.
Consequently, new information from locked chats will not leak into general conversations. Resetting a PIN permanently deletes all locked chats, files, and other content protected by that previous code.
The option is unavailable for temporary chats, archived chats, group chats, links shared with others, chats containing scheduled tasks, and meetings with recordings.
Availability
Locked Chats is available on ChatGPT Free, Go, Plus, Pro, Business, Enterprise, and Edu plans. It is not available on ChatGPT for Healthcare. The feature began rolling out on Tuesday and may not be live for all users yet.
What it means
People making things now have a way to stop others from seeing specific conversations on their device without deleting the data. The lock protects the view, but it does not protect the data from being stored or used by the model in the background.
