We’ve all been there: you are sharing your screen during a work presentation, or handing your phone to a friend to show them a cool photo, and a stray ChatGPT history sidebar pops up revealing personal brainstorming sessions, private thoughts, or sensitive prompts. While ChatGPT has long been our go-to assistant for everything from professional coding to personal reflection, keeping those intimate interactions hidden from casual onlookers has always required manually deleting chats or digging through privacy settings.
OpenAI is finally solving this friction with a major new privacy upgrade: Locked Chats.
Designed to keep your most confidential conversations completely shielded from prying eyes, this feature brings a dedicated, secure layer to your account across almost every subscription tier. But how does it work, what are the nuances behind its memory system, and does a simple PIN truly keep your data safe? Let’s break down everything you need to know.
How ChatGPT Locked Chats Work
The new Locked Chats feature acts as a secure digital safe within your ChatGPT interface. Located directly under Projects > Locked Chats, it allows you to segregate your sensitive conversations away from your main workspace into an isolated, protected zone.
Here is what happens when you lock a conversation:
Complete Sidebar Isolation: Moved chats are hidden from your regular sidebar, pinned lists, and standard global search results.
Multi-Device Authentication: To open your locked section, you authenticate using an account 4-digit PIN on web and desktop apps, or biometrics (Face ID, Touch ID, or device unlock) on iOS and Android.
Auto-Locking Mechanisms: Access is temporary. On web and desktop, sessions relock when navigating away or after 1 minute in the background. Mobile apps automatically relock after 1 minute of inactivity or when minimized, with an option to tap “Lock now” for instant protection.
Protected Files and Attachments: Any documents, code files, or images generated or uploaded within a locked chat are shielded and remain completely hidden from your broader file space and un-locked chats.
Best of all, OpenAI isn’t locking this behind an expensive paywall. The feature is rolling out across Free, Plus, Pro, Business, Enterprise, and Edu plans, ensuring everyone from casual users to corporate teams can benefit from enhanced physical device privacy.
How Locked Chats Handle Memory and Shared Links
One of the most critical updates involves how OpenAI handles memory within locked sessions:
Isolated Memory System: If ChatGPT’s Memory feature is enabled on your account, Locked Chats utilizes its own dedicated, separate memory system. Information gathered while conversing inside locked chats can be saved to this isolated memory without bleeding into your regular everyday interactions.
No Retroactive Erasure: Locking an existing conversation does not delete facts or context that ChatGPT already saved to your regular memory before the lock was applied. To prevent leakages, security experts recommend starting sensitive conversations directly inside the Locked Chats folder rather than locking them after the fact.
Existing Shared Links: If a chat already has an active public or team share link, locking the conversation will not automatically revoke that link. While new messages sent inside the locked state won’t be appended to the public link, anyone with the existing link can still view historical messages unless you manually delete the shared link.
Locked Chats vs. Temporary Chats vs. Deletion
To help you choose the right privacy tool for your workflow, here is a quick breakdown of how Locked Chats compare to ChatGPT’s other privacy modes:
| Feature / Mode | Appears in Sidebar? | Keeps Conversation History? | Memory System Used | Protects Against Local Onlookers? |
| Normal Chat | Yes | Yes (Saved until deleted) | Regular Memory | No |
| Locked Chat | Only inside Locked Chats folder | Yes (Saved and PIN-protected) | Separate Locked Memory | Yes (Requires PIN / Biometrics) |
| Temporary Chat | No | No (Wiped after session) | Disabled (No memory saved) | Partial (Hidden while active) |
| Deleted Chat | No | Permanently Removed | Requires manual removal | Yes |
The Fine Print: What Locked Chats Do Not Protect
While Locked Chats are a massive leap forward for physical device security and screen-sharing safety, it is vital to understand what the feature does not do:
No End-to-End Encryption (E2EE): Locked Chats are protected by an app-level UI lock, not end-to-end zero-knowledge encryption. OpenAI still stores and processes the data on its servers according to your account’s data retention policies.
Unchanged Data Training Settings: Locking a chat does not automatically opt you out of model training. If your account allows OpenAI to train on your conversations, locked chats follow the exact same global setting unless manually turned off in settings.
The Forgotten PIN Danger: Resetting your 4-digit PIN is a destructive operation. If you forget your PIN, resetting it will permanently wipe all locked chats, dedicated memory, and associated files. OpenAI customer support cannot recover wiped data.
Is a PIN Enough? The Final Verdict
For professionals presenting decks in boardrooms, students sharing laptops, or individuals who value a clear boundary between work and personal reflections, Locked Chats offer invaluable peace of mind. You no longer have to delete helpful prompt histories just to keep your workspace presentable.
While it isn’t a replacement for zero-trust encryption or strict data hygiene, it serves its primary purpose exceptionally well: preventing local visual shoulder-surfing.
What do you think? Is a 4-digit PIN and Face ID protection enough for your day-to-day AI chats, or are you holding out for client-side encryption?
For continuing updates on OpenAI, ChatGPT, Codex, GPT models, and AI industry developments, bookmark:
