ChatGPT now lets you lock your chats: The company says the feature is

OpenAI rolled out a dedicated chat-locking security feature on Tuesday that allows users to protect individual conversations behind authentication on shared screens and mobile hardware. According to TechCrunch's report, the update introduces hardware-level and passcode barriers directly within the application, giving users the ability to restrict conversation visibility using a four-digit PIN or native biometrics such as Face ID and Touch ID on supported devices. The mechanism addresses everyday screen-sharing scenarios where people hand their unlocked phones or laptops to relatives, friends, or coworkers without wanting sensitive session histories exposed.

This breakdown details the operational mechanics of the new security tier, the strict boundaries of its local privacy protections, its isolated memory architecture, and its tiered commercial availability. It is written for everyday consumers, workplace professionals, and system administrators who handle sensitive data on OpenAI client applications and need to know exactly which conversation types, subscription tiers, and storage systems fall within or outside this new privacy perimeter.

ChatGPT now lets you lock your chats: what actually changed

The most immediate change is the removal of protected conversations from the standard interface components that users navigate every day. Once a user locks an active discussion, the thread vanishes from regular chronological chat history, pinned conversation lists, and standard chat search queries. Instead, the dialogue transfers into a dedicated Locked Chats section that demands authentication before any titles, historical turns, or uploaded documents become readable on the display. This structure prevents casual observers or someone inspecting recent search queries from stumbling upon sensitive materials.

Unlike the existing Temporary Chats mechanism, which discards sessions entirely and prevents them from ever appearing in account logs, locked conversations remain permanently accessible to the account holder. Users can step away from a session, secure the application, and unlock the conversation at any point in the future to continue querying the model. The feature functions as a persistent archive behind an authentication wall rather than an ephemeral burner session, giving users continuity without sacrificing basic visual confidentiality.

ChatGPT now lets you lock your chats: how it works

Securing a conversation requires setting up an authentication method directly tied to the client application and supported device. Users configure a four-digit PIN or authorize hardware biometric access through Touch ID or Face ID on compatible mobile phones and computers. Entering the correct credentials unlocks the isolated section, revealing the saved conversations. If a user forgets their passcode or attempts an administrative reset, OpenAI enforces a hard data boundary: resetting the four-digit PIN permanently wipes every locked conversation, attached file, and associated asset protected under that previous code.

The underlying memory subsystem also bifurcates as soon as an active thread enters the protected state. When account-level Memory remains enabled, the Locked Chats partition creates an isolated, internal memory pool shared exclusively among conversations inside that section. Incoming messages sent after locking can update this compartmentalized store, but they are barred from updating regular account memory. Consequently, personal facts, names, or project details disclosed inside a locked interaction will not leak or surface in ordinary, unlocked chats.

ChatGPT now lets you lock your chats: why it matters now

OpenAI developed the capability because users increasingly treat conversational artificial intelligence as an intimate personal diary, financial planner, health sounding board, and gift organizer. The company specifically designed the feature for common social interactions, such as passing a handset to a friend to view a recipe or work presentation, where adjacent personal chats might otherwise be exposed in the sidebar. Introducing an on-screen barrier allows users to share their devices casually without exposing unrelated private discussions taking place on the same account.

At the same time, the capability carries clear architectural limits that users must understand before treating the tool as a corporate safe. OpenAI explicitly clarified that locking a conversation does not establish end-to-end encryption or route files into a cryptographic vault. The retention schedule governing stored conversations remains completely unchanged, and locking an exchange does not delete contextual facts remembered by the model before the lock was initiated. Furthermore, the feature does not modify the existing Improve the model for everyone training toggle.

ChatGPT now lets you lock your chats: who is affected

The rollout touches almost the entire consumer and workplace portfolio across global platforms. OpenAI confirmed that Locked Chats will be available to members on ChatGPT Free, Go, Plus, Pro, Business, Enterprise, and Edu subscription tiers. Because the deployment started on Tuesday, access is phasing across the network and may take time to populate every client installation. Accounts managed across teams and academic institutions can leverage the authentication barrier on employee and student hardware alongside individual retail accounts.

However, OpenAI established notable exclusions across specific enterprise domains and message formats. ChatGPT for Healthcare accounts do not receive the feature under this release. Furthermore, users cannot apply locks to temporary chats, archived chats, existing group chats, conversations launched from shared web links, automated chats bound to scheduled tasks, or sessions containing meeting recordings. Any workflow relying on those excluded operational modalities must rely on alternate administrative controls or local device locks.

ChatGPT now lets you lock your chats: what to watch

Users should closely monitor how their workflow handles information recorded prior to applying a passcode. Because locking a conversation only isolates memory generated from new messages sent after the lock, any profile detail ingested prior to that moment remains resident in general memory. Account owners managing sensitive plans must review their preexisting memory logs manually to confirm that older personal disclosures are not recalled by the assistant during routine, unlocked daytime queries on the same profile.

Device management and passcode hygiene also require strict operational care due to the unrecoverable nature of the security design. Because OpenAI engineered the four-digit PIN reset as an absolute purge, users cannot request support tickets or cloud restorations to retrieve wiped data after credential loss. Individuals storing important project notes or personal archives inside the protected section must ensure their four-digit code is safely remembered, or accept that resetting their access credentials will instantly and permanently erase every locked record.