OpenAI adds locked chats to ChatGPT as privacy pressure grows

OpenAI has introduced a new “Locked Chats” option in ChatGPT, giving users a way to shield sensitive conversations with a PIN or device biometrics when they need to hand over their phone, tablet, or computer. The feature matters because ChatGPT is increasingly used for private planning, personal questions, and sensitive brainstorming, making conversation privacy a bigger concern for everyday users.

The new feature, which began rolling out on Tuesday, creates a separate protected space for selected chats. Users can lock conversations with a four-digit PIN or, on supported devices, unlock them with Face ID or Touch ID. Unlike Temporary Chats, locked chats remain available later and can be revisited, while still staying out of the regular chat list unless the user chooses otherwise.

What OpenAI changed and why it matters

OpenAI’s update gives ChatGPT users a more practical privacy option for moments when a conversation is personal but not meant to be seen by someone looking over their shoulder.

The company framed the feature around ordinary, real-world use cases: a user may be discussing a surprise gift, arranging a private event, or exploring something sensitive and then need to show a friend or family member a different chat. In those situations, a lock is simpler than deleting a conversation or starting over in a temporary session.

That distinction is important. Temporary Chats are designed to avoid appearing in history from the start, while Locked Chats preserve the conversation for later access. In other words, OpenAI is trying to address privacy at the device level without making users sacrifice continuity.

How Locked Chats work in ChatGPT

Locked Chats create a dedicated section inside the app where protected conversations are separated from normal history, pinned chats, and search results. When a user locks a conversation, it moves out of the standard view and into that restricted area.

Access is controlled in one of two ways: users can enter a four-digit PIN, or they can rely on Face ID or Touch ID on compatible devices. The feature is intended to be quick to use, so users do not need to set up a separate account or external password manager to protect a single conversation.

What happens to Memory in locked chats?

Locked Chats get their own memory context when ChatGPT Memory is turned on. New information added in that protected section can shape future locked conversations, but it does not spill into regular chats. OpenAI says this is meant to keep sensitive details compartmentalized rather than merged into a user’s broader ChatGPT profile.

That separation is one of the more notable design choices in the update. It suggests the company is trying to reduce accidental crossover between private discussions and everyday interactions with the assistant.

How is this different from a secure vault?

It is not the same as a fully encrypted private vault. OpenAI says locking a chat does not create end-to-end encryption, does not change how long the chat is retained, and does not remove information ChatGPT may already have remembered before the chat was locked. It also does not affect the user’s “Improve the model for everyone” setting.

That means the new feature is best understood as an access-control tool, not a comprehensive privacy overhaul.

OpenAI says the feature is aimed at everyday situations where users want to keep a chat out of sight from people nearby, especially when they are handing over a device or showing someone something else on-screen.

What Locked Chats do and do not protect

Although the feature raises the privacy bar, it has clear limits. Users should not assume that a locked conversation is invisible to all systems, or that it changes the way OpenAI stores data behind the scenes.

OpenAI’s own explanation makes the boundaries explicit: the lock hides chats from casual access and separates them from regular history, but it does not create a private encrypted container with all the protections that phrase might imply.

The distinction matters because consumers often assume “locked” means “fully private.” In practice, OpenAI is offering a usability feature that limits who can open a chat on a shared device, not a complete redesign of the company’s data practices.

Why users may want a chat lock

Many people now use ChatGPT for far more than casual prompts. The assistant is often a place to draft messages, plan travel, search for advice, organize finances, prepare gift ideas, or think through personal issues. Those are exactly the kinds of conversations that may feel awkward or risky if someone else picks up the device.

The lock feature is meant to reduce that friction. Instead of deleting a thread, users can keep it around while making it harder to stumble across. That is especially useful in households, offices, classrooms, and shared workspaces where devices are often borrowed or passed around.

• Planning a surprise without spoiling it
• Saving personal advice for later review
• Keeping sensitive notes off the main chat list
• Preventing accidental viewing when showing someone another conversation

Which chats can be locked?

Not every ChatGPT conversation is eligible for the new feature. OpenAI says Locked Chats are not available for temporary chats, archived chats, group chats, chats created from a shared link, chats with scheduled tasks, or chats with meeting recordings.

That limitation narrows the feature’s scope, but it also helps explain its purpose. OpenAI appears to be prioritizing one-on-one user chats that are most likely to contain personal information or private planning.

Availability across plans

Locked Chats are rolling out to users on ChatGPT Free, Go, Plus, Pro, Business, Enterprise, and Edu plans. OpenAI says the feature is not available in ChatGPT for Healthcare.

Because the rollout started on Tuesday, some users may not see the option immediately. As with many product launches, availability can vary by account, region, app version, or staged release timing.

How does this affect ChatGPT privacy more broadly?

Locked Chats improve everyday privacy, but they do not resolve broader concerns about how AI platforms store, use, and remember user information. That is the central tension in this release.

On one hand, the feature gives users a more convenient way to keep conversations out of casual view. On the other, OpenAI is careful to say that locking a thread does not change the underlying handling of data or retroactively erase memory associated with it.

That means the update should be seen as incremental rather than transformative. It is a useful shield for shared screens and shared devices, but it does not answer every question about confidentiality, retention, or model training.

For privacy-conscious users, the practical takeaway is straightforward: if your concern is someone physically seeing a conversation on your device, Locked Chats helps. If your concern is deeper data governance, the feature does not substitute for a full privacy policy review or a separate secure workflow.

Why OpenAI is adding more privacy tools now

ChatGPT has moved beyond novelty and into daily utility, which means it increasingly handles material people would once have written in a notes app, sent in a private message, or kept offline. As usage becomes more personal, privacy expectations rise with it.

That shift helps explain why OpenAI is introducing more nuanced controls rather than relying only on broad account settings. The company needs features that fit real-life behavior, especially on devices that are regularly shared among partners, family members, students, and coworkers.

In that sense, Locked Chats is part of a broader trend across consumer software: privacy features are becoming more contextual, more granular, and more focused on everyday convenience.

What users should know before relying on Locked Chats

Users should treat the feature as a visibility control, not as a legal-grade or security-grade safeguard for especially sensitive data.

1. It protects access on the device, not every possible backend system.
2. It does not rewrite ChatGPT’s retention or memory rules.
3. It is not available for every kind of chat.
4. It may not appear immediately while rollout continues.

For most people, that will still be enough for the feature to be genuinely useful. The most common privacy problem with AI chats is not always sophisticated hacking; it is often simple exposure through an unlocked screen or a shared device. OpenAI’s update is aimed squarely at that problem.

Bottom line

OpenAI’s Locked Chats feature gives ChatGPT users a straightforward way to hide selected conversations behind a PIN or biometrics and keep them separate from normal history. It is a practical privacy upgrade for everyday use, especially on shared devices, but it stops short of offering encrypted vault-level protection.

As ChatGPT becomes a place for more personal and sensitive conversations, features like this are likely to become less of a nice-to-have and more of a baseline expectation.