ChatGPT adds Locked Chats secured by PIN, Face ID or Touch ID

OpenAI announced on Tuesday that ChatGPT users can now lock individual conversations behind a four-digit PIN, Face ID or Touch ID. The feature, called Locked Chats, is starting to roll out across nearly every ChatGPT plan, from Free to Enterprise.

The use case is deliberately mundane. You hand your phone to a friend or family member to show them something in one chat, and you don't want them scrolling into another. OpenAI says the feature targets situations where users discuss something personal, or plan a surprise, and want that conversation kept private while other people have access to their screen or device.

That framing matters. ChatGPT is no longer just a work tool or a coding assistant. People increasingly use it for personal conversations, and personal conversations create a new category of privacy problem: one that lives on the device, in the chat history, visible to anyone holding the phone.

What exactly does locking a chat do?

Once a user locks a chat, it moves into a dedicated "Locked Chats" section. That section sits apart from the regular chat history, apart from pinned conversations, and apart from chat search results. The conversation doesn't disappear; it just requires authentication to open again.

Users can secure locked chats in one of two ways:

• A four-digit PIN
• Face ID or Touch ID, on supported devices

And unlike Temporary Chats — the app's existing privacy feature that runs a separate session and never appears in chat history — Locked Chats persist. Users can return to them at any time. Temporary Chats are ephemeral by design. Locked Chats are permanent conversations with a lock on the door.

How does Locked Chats handle Memory?

This is where the feature gets more technically interesting. With Memory enabled, OpenAI says the Locked Chats section has its own separate memory, shared across conversations within that section but walled off from the regular Memory feature.

New messages sent after a chat is locked can update the locked-section memory. They do not update regular Memory. In OpenAI's words, that means new information from locked chats won't leak into regular chats.

That separation cuts both ways, and users should understand the boundary. Locking a chat does not scrub information that ChatGPT remembered before the chat was locked. Anything the assistant already absorbed into regular Memory stays there. The lock protects the conversation going forward, not the model's recollection of the past.

What doesn't the lock protect?

OpenAI is unusually direct about the limits here, and the caveats deserve attention. According to the company, locking a chat:

• Does not create end-to-end encryption or a separate encrypted vault
• Does not change how long OpenAI retains the chat
• Does not remove information ChatGPT remembered before the chat was locked
• Does not change the "Improve the model for everyone" setting

That last point carries the most weight. If the "Improve the model for everyone" setting is active, the content of a locked chat can still feed model training, lock or no lock. The feature is a privacy control for the device and the interface — for who can open the app and read the thread — not a privacy control for the data pipeline behind it.

Locking a chat limits who sees the conversation. It does nothing about how the information inside it is stored or handled on OpenAI's side.

What happens if you reset the PIN?

There is one hard-edged consequence built into the design. OpenAI says resetting a PIN permanently deletes all locked chats, files, and other content protected by the previous PIN.

Forget the code, and the recovery path is deletion, not restoration. The chats don't unlock; they vanish, along with anything attached to them. Users who rely on Locked Chats to store sensitive files should treat the PIN as load-bearing.

Which chats and plans are covered?

Locked Chats doesn't apply to everything in the app. OpenAI lists several exclusions, and the feature is unavailable for:

• Temporary chats
• Archived chats
• Group chats
• Chats created from a shared link
• Chats with scheduled tasks
• Chats with meeting recordings

On the plan side, coverage is broad. OpenAI says Locked Chats will be available to users on ChatGPT Free, Go, Plus, Pro, Business, Enterprise, and Edu plans. The one exception is ChatGPT for Healthcare, where the feature is not available — a notable gap given that health-related conversations are arguably among the most sensitive a user could want to lock.

The rollout started Tuesday, so the feature may not be live yet for all users.

Why this matters beyond the PIN pad

Locked Chats is a small feature with a telling shape. It acknowledges that ChatGPT now hosts conversations people consider genuinely private — the kind worth hiding from a partner, a roommate, or a curious family member borrowing the phone.

But the design also draws a bright line between interface privacy and data privacy. The lock stops someone standing next to you from reading your chat. It doesn't encrypt anything, doesn't shorten retention, doesn't touch training settings, and doesn't erase what the model already remembers. For users weighing whether to move sensitive discussions into ChatGPT, that distinction is the one to keep in view — and the rollout's gradual arrival means most users will meet the feature in the wild before they read the fine print.