In this story ⌄
- Locking changes access and visibility, not your training or retention settings.
- Resetting a forgotten account PIN permanently deletes the content protected only by that PIN.
A chat title can reveal plenty before anyone opens the conversation. That is an awkward discovery halfway through sharing your screen.
OpenAI’s October 20 release notes introduced locked chats: selected conversations move out of regular history, pinned chats and search, with an account PIN or supported device authentication used to open them. The rollout is gradual, so the option may not have reached your account.
Useful? Certainly as a way to make a sidebar less revealing. But a padlock icon can suggest more than the feature actually does.
Three settings, three different jobs
OpenAI’s data-controls guide separates saving a conversation from allowing it to help train models. Turning off Improve the model for everyone stops new conversations being used for training while letting you keep your chat history. Locking a saved chat leaves that training preference in place.
Temporary chat is another option entirely: while temporary, those conversations stay out of history, do not create or update memories, and are not used to improve models. OpenAI says they may still be kept for up to 30 days for safety. Saving one changes how it is treated.
Think about the outcome you want before picking the setting. Keeping a conversation available to revisit, keeping its title out of view, and controlling training use are separate decisions. One convenient icon cannot express all three.
Read the PIN warning
The locked-chat documentation says locking does not add end-to-end encryption, change retention or remove previously remembered information. Existing shared links also survive; delete a link separately to stop access to content already shared through it.
There is a sharper catch with recovery. Resetting a forgotten account PIN permanently deletes locked chats and the information or files protected only by that PIN. OpenAI says it cannot recover that content. This is different from changing a PIN you still know.
Our practical advice is to read that warning before treating reset as a routine troubleshooting button. Start with a conversation you can afford to experiment with, and check what remains visible before your next presentation.
A privacy control earns its place when you know exactly which problem it solves. The sidebar was one problem. The rest deserve their own settings.
Sources checked October 23, 2026.
Understand Temporary chat
Temporary chat has its own controls, shown here in OpenAI’s mobile help image. This is not the Locked chats screen.
Try this next.
A few small steps to turn the idea into something useful.
Go a little deeper.
Original sources, practical instructions, and places worth exploring.
ChatGPT release notes
October 20 announcement and gradual rollout of locked chats.
help.openai.comUsing locked chats in ChatGPT
Access controls, existing shared links and the consequences of resetting a PIN.
help.openai.comData controls in ChatGPT
How model-training preferences and temporary chats differ from hiding a saved conversation.
help.openai.comExternal links open in a new tab. References and further reading are selected for this story; inclusion is not an endorsement.
Something we should know? Send a tip or correction.
Your take.
A different angle. A good question. A hill you’ll happily die on. We’re listening.
Be the first to add a little signal.



