ChatGPT Can Lock Private Chats Behind PIN, But With No Recovery Option 

OpenAI began rolling out locked chats in its ChatGPT on Tuesday, October 20, 2026, that lets users pull selected conversations out of their regular history and open them only with a four-digit account PIN. 

On iPhone and Android, Face ID, Touch ID, or device unlock can stand in for the PIN. 

The feature runs on the web, in the mobile apps, and in the new desktop app for macOS and Windows. OpenAI built it for moments when a screen is visible to others, such as sharing a display or handing a phone to a relative.

What OpenAI’s Locked Chats Actually Hide

Users can lock an existing conversation by selecting the lock option from its three-dot menu, choosing Lock chat, and setting a four-digit PIN. 

Alternatively, you can start a private thread from scratch by going directly to Projects > Locked chats and launching a new conversation there. 

Once locked, the chat moves out of your main history, pinned list, and search results, keeping its title hidden in the sidebar until you authenticate. 

Access is brief by design. On the web, refreshing or closing your tab locks the chat again. On iOS and Android, it locks after one minute in the background or away from the screen.

Not every conversation can be locked. Temporary chats, archived chats, group chats, tasks tied to schedules or meeting recordings, shared link creations, and Codex threads are all excluded. 

Locking a chat inside a project pulls it out, meaning it loses that project’s instructions, files, and memory. 

One caveat matters during a screen share: the AI lab says other open devices may keep showing a chat’s title until their lists refresh.

How ChatGPT Memory, Files and Admins Treat Locked Chats

With Memory on, a locked chat can still draw on what ChatGPT already knows about you, and nothing remembered before the lock is erased. 

What it learns afterwards stays inside the locked area and never feeds regular chats, which TechCrunch describes as a separate memory shared across locked conversations.

Files follow the chat. Anything uploaded or created inside it is hidden from Space, while files pulled in from elsewhere stay visible. A shared link also survives the lock, so previously shared content remains reachable until you delete the link yourself.

Workspace rules still apply. Administrators cannot unlock chats, reset a PIN, or switch the feature off, yet Enterprise and Edu locked chats still appear in message logs on OpenAI’s Compliance Platform.

Why OpenAI’s PIN Reset Rule Deserves a Second Look

A simple four-digit PIN only offers 10,000 combinations, so OpenAI enforces temporary lockouts if you guess wrong too many times. Forget that PIN, though, and every locked chat and its attached files are wiped permanently. 

OpenAI Help Center, which covers all about the Locked chats, makes it clear that support cannot bring them back.

That design makes the feature honest. It shields your screen from nearby onlookers, not from OpenAI. 

Data retention, AI model training opt-outs, and ads work exactly as they did before, and anyone with access to an open chat can still copy its text.

If you want stricter data boundaries, Temporary Chat is better suited because it creates no memories and stays out of model training. 

It just cannot be reopened later. Locked chats work well for short-term privacy, like planning a surprise party, but anything you do not want OpenAI to hold does not belong in either.

Source: Using locked chats in ChatGPT