ChatGPT locked chats: PIN privacy, not encryption
OpenAI is rolling out a PIN- and Face ID-protected “locked chats” space for ChatGPT. It adds visibility controls for sensitive work but remains a retention-linked product feature, not end-to-end encryption.
What happened
OpenAI has begun rolling out a “locked chats” feature for ChatGPT, designed to keep sensitive conversations out of the main chat history, pinned chats and conversation search. Users who enable it access those conversations through a 4-digit PIN, while iPhone and Android users can also use authentication systems like Face ID. Locked conversations are grouped under a project titled “Locked chats.”
The feature is arriving gradually; OpenAI says it may take a few days to appear in an account. It will cover free accounts as well as Go, Plus and Pro subscriptions. On Android and iOS, voice mode can be used with locked chats, but Codex conversations cannot be locked.
OpenAI is explicit about the limits: locked chats are not end-to-end encryption and not a way to avoid legal obligations. A support page notes that the feature will not automatically change model training settings or how long OpenAI retains conversations. The company also warns that resetting an account PIN deletes all locked chats and any files uploaded to their project space. “This cannot be undone, and OpenAI Support cannot recover the deleted content,” OpenAI states.
Key facts
- Locked chats can be kept out of regular chat history, pinned chats and conversation search; access requires a 4-digit PIN or, on iPhone and Android, authentication systems like Face ID.
- Private conversations are grouped under a project titled “Locked chats,” but this is not end-to-end encryption and not a legal shield.
- Locked chats will not automatically change model training settings or how long OpenAI retains private conversations.
- A PIN reset deletes all locked chats and uploaded files in the project space, and OpenAI Support cannot recover the content.
- Rollout is gradual and covers free, Go, Plus and Pro accounts; voice mode works with locked chats on Android and iOS, but Codex cannot be locked.
Our analysis
For operators and creators, this changes visibility rather than security. A ChatGPT thread containing an unreleased campaign, sensitive client feedback or draft community language can now be moved out of search and regular history, reducing the chance that a colleague or someone handling your phone sees it. That control is most relevant for shared devices, public workspaces and situations where a phone is handed to someone else.
The retention detail is the more important risk. Because locked chats will not automatically change model training settings or retention periods, users who assume “locked” means private in the encrypted sense are likely misreading the tool. OpenAI frames the feature mainly as a way to prevent people nearby from seeing conversations, not as end-to-end encryption. For social media managers and community operators handling brand or user data, this suggests locked chats should be treated as a screen-level privacy layer, not a compliance boundary.
The PIN reset rule is likely to create a specific operational hazard. If a user forgets a PIN and resets it, all locked chats and files in that project space are deleted with no OpenAI recovery. Anyone storing high-value prompts, content drafts or client-approved copy only inside locked chats has a single point of failure. Teams should likely assume locked chats can disappear and keep an independent record of anything critical.
What it means for operators
- Use locked chats for pre-launch scripts, client-sensitive copy and personal notes that should not appear in regular history, pinned chats or conversation search.
- Do not treat locked chats as end-to-end encryption or a legal shield; keep regulated or legally sensitive material in systems built for that purpose.
- Because a PIN reset deletes all locked chats and uploaded files, store a separate copy of high-value outputs outside ChatGPT.
- Check model training and retention settings independently; locked chats will not automatically change them.
- For mobile workflows, lock conversations with authentication systems like Face ID, and remember that voice mode works on Android and iOS, but Codex chats cannot be locked.
Source:ChatGPT now lets you lock sensitive chats behind a PIN — Engadget(2026-10-20)
Editor's note: prepared by the NoobClaw newsroom with AI assistance from the public report above. Facts are as reported by the source; the analysis is our view. Spotted an error? Contact us and we will correct it.