ChatGPT Launches Locked Chats with PIN and Biometric Authentication

OpenAI announced on Tuesday (October 20) that ChatGPT users can now lock individual conversations with a four-digit PIN or biometric authentication via Face ID and Touch ID. The feature, called "Locked Chats," is designed to protect sensitive conversations when other people may have access to the user's screen or device.

The feature had been requested by the OpenAI community for months. In June 2026, a user on OpenAI's feature requests community specifically asked the company to add the ability to lock individual conversations with Face ID, Touch ID, or a PIN, moving them into a protected folder. The OpenAI support team's response at the time indicated that while the feature made sense, the company had no timeline — which proved prescient, as the functionality launched exactly four months later, in October 2026.

How the locking works

Once activated, ChatGPT creates a dedicated section called "Locked Chats" in the app interface. Locked conversations are kept separate from the regular chat history, pinned conversations, and chat search results. This means that when someone looks at the user's device screen, they won't see the content of locked conversations — only that they exist.

If the user has the "Memory" feature enabled, locked conversations maintain their own separate memory, shared among conversations within that section. Importantly: new information entered in locked conversations updates only the isolated memory, not the regular memory. In other words, ChatGPT does not use information from locked conversations to personalize responses in normal chats. This separation is a relevant technical detail that many users may not immediately notice.

What locking does NOT do

OpenAI was explicit in its caveats: locking a conversation does not create end-to-end encryption, nor does it generate a separate encrypted vault. The feature does not alter the data retention period by OpenAI, does not remove information that ChatGPT had already memorized before the conversation was locked, and does not modify the "Improve the model for everyone" setting. In other words, the lock protects against accidental screen glances — not against data collection and storage by OpenAI's servers.

This distinction is fundamental. The feature is a local privacy protection (on the user's device), not a global privacy protection (against data collection and storage). Users seeking complete anonymity in conversations with AI will still need to resort to more drastic methods, such as using the temporary chat mode (which, notably, does not qualify for the locking feature).

Plans and limitations

The feature is available across all ChatGPT plans: Free, Go, Plus, Pro, Business, Enterprise, and Edu. The only exception is ChatGPT for Healthcare, which already has its own regulatory compliance mechanisms. The feature began rolling out on Tuesday (October 20), and OpenAI stated it may not be available for all users immediately.

The exclusions are numerous: locking is not available for temporary, archived, group, shared-link, scheduled-task, or meeting-recording conversations. The most surprising restriction might be the inability to lock group conversations — a plausible use case for professionals who share devices with colleagues.

Resetting the PIN erases everything

One detail that deserves attention: OpenAI states that resetting the PIN permanently deletes all locked conversations, files, and other content protected by the previous PIN. There is no recovery or backup option. For users who switch devices or forget their PIN, this means irreversible data loss.

The broader context

The arrival of locked chats reflects ChatGPT's evolution from a productivity tool to a personal conversation space. As the app is increasingly used for journals, medical advice, event planning, and even informal therapy, the need for privacy has become more acute. The feature, while limited, is a step in that direction — but it does not resolve the broader concerns about how OpenAI stores and uses sensitive conversations.

While the company does not introduce end-to-end encryption (a feature many users have requested and that Anthropic already offers in Claude), local locking is the best we have for protecting conversations on-device. Whether OpenAI will eventually extend this to real encryption, or whether local locking will be the privacy ceiling the company offers — and whether that will be enough for the millions of users who already trust ChatGPT with their most personal secrets — remains to be seen.

Sources: TechCrunch, PCMag, Firstpost

✓ Independent sources cross-checked and verified before publishing