ChatGPT Now Locks Chats Behind a PIN, but Forgetting It Deletes Them Forever

OpenAI began rolling out a Locked Chats feature in ChatGPT on Tuesday, letting users hide individual conversations behind a four-digit account PIN or their phone biometrics. The feature is available across every plan from Free and Go to Plus, Pro, Business, Enterprise and Edu, on the web, iOS and Android apps, and the new desktop applications for macOS and Windows. ChatGPT for Healthcare is not included, and the rollout is gradual, so the option may take a few days to appear in some accounts.

Using it is simple: open the conversation, tap the three-dot menu, choose Lock chat, and set the PIN the first time. Locked conversations disappear from regular history, pinned chats and search, and live in a dedicated section under Projects called Locked chats. On iPhones and Android phones users can additionally enable Face ID, Touch ID or device unlock, though OpenAI is explicit that device authentication is optional, applies only to that device, and never replaces the PIN, which belongs to the account rather than any one phone.

Access windows are deliberately short. On mobile, locked chats relock after about one minute away from the conversation or the app, and on the web they relock when you leave the tab, refresh or close it. Not everything can be locked: temporary chats, archived chats, group chats, conversations opened from shared links, chats with scheduled tasks or meeting recordings, and Codex conversations are all excluded.

OpenAI is unusually blunt about what the feature does not do. Locking adds no end-to-end encryption and no separate encrypted vault, changes nothing about how long OpenAI retains the chat, does not alter the Improve the model for everyone setting, and does not override workspace policies or legal obligations. Anyone who can open a locked chat can still copy its contents. The company frames the feature as protection against the person next to you, a shared laptop, a projector in a meeting, not against OpenAI itself.

The sharpest edge is the recovery story. Changing the PIN while you still know the current one is safe and keeps everything. But resetting a forgotten PIN permanently deletes every locked conversation, along with memories and files that existed only inside them, and OpenAI says support cannot recover any of it. Memory is also segregated: information remembered inside locked chats stays there and does not flow into regular conversations, while anything remembered before locking remains.

For organizations there is a notable power dynamic. Business, Enterprise and Edu admins cannot disable the feature, cannot unlock a chat, and cannot reset an employee PIN, yet locked chats still appear in the Enterprise compliance platform message logs, so workplace compliance exports keep capturing them. Employees should assume a locked chat is hidden from colleagues, not from IT.

The feature lands as consumer AI privacy expectations tighten across the industry, and as ChatGPT turns into the place people discuss health worries, job searches and finances. A shoulder-surfing guard was an obvious gap, and the destructive PIN reset is the price OpenAI chose for keeping the locked area truly keyed to memory rather than an email reset link.