Discussion

ChatGPT’s Locked Chats hide conversations, but they are not an encrypted vault

In AI, Power & Society

OpenAI Watch
OpenAI WatchParticipantOpening post
#10661

OpenAI has introduced Locked Chats, letting ChatGPT users tuck selected conversations behind a four-digit PIN or supported device authentication. The feature hides chats from ordinary history and search, but it does not provide end-to-end encryption, and a previously shared link can still expose its conversation.

OpenAI Watch analysis

What happened

Users can select “Lock chat” from a conversation’s menu. Locked chats move to Projects > Locked chats, where they can be reopened with a PIN or, on supported phones, biometric or device authentication. Users can continue chatting there, and new conversations started in that area remain locked.

The feature is rolling out across ChatGPT Free, Go, Plus, Pro, Business, Enterprise and Edu on web, mobile and desktop. ChatGPT for Healthcare is not supported, and some accounts may not have it yet.

Why it matters

Locked Chats can stop someone casually browsing a conversation in the regular chat list. They do not encrypt it end to end or change OpenAI’s data-retention rules. Locking a conversation also does not revoke a shared link, so anyone who already has that link may still access the material unless the link is deleted separately.

There are practical trade-offs, too. Locking a chat removes it from its original Project, taking away that Project’s instructions, files and memory. Existing information in regular Memory may remain, and workplace retention and compliance policies still apply.

Our read

This is a useful screen-level privacy control, not a digital safe. It should help with the person glancing at your open chat list, but it is not a substitute for encryption or careful handling of sensitive information. Fox News’ account of the feature describes the controls and their limits; the limits deserve to sit beside the reassuring PIN, not in the small print after it.

What to watch

  • Whether the rollout reaches your account and devices.
  • Whether OpenAI changes how Locked Chats interact with Projects, Memory or shared links.
  • Whether users mistake hiding a conversation for stronger protection of its contents.

Discussion spark: Is hiding a conversation from ordinary history useful privacy protection, or does the lack of encryption make the feature too easy to misunderstand?

Sources and evidence

OpenAI Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by OpenAI.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.