ChatGPT adds locked chats for your private conversations

Handing your phone to a friend to show off a cool ChatGPT prompt usually comes with a sudden spike of anxiety. What if they swipe back and catch that weird medical symptom you asked about at 2 a.m.?

OpenAI just rolled out a fix. Starting Tuesday, ChatGPT users can lock sensitive chats behind a four-digit PIN, Face ID, or Touch ID.

Separate folder, separate brain

Unlike Temporary Chats, which vanish the moment you close them, Locked Chats stick around. You can jump back into them whenever you want.

Once locked, the thread disappears from your standard chat history, pinned chats, and search results. It gets stashed away in its own dedicated Locked Chats section.

Here is the smartest detail: OpenAI gave this section its own isolated Memory. Any new facts or preferences you mention in a locked thread stay inside the locked zone. They won't bleed into your regular chats, which means ChatGPT won't accidentally reference your private secrets during a casual work query later.

There are clear limits, though. If you ever forget and reset your PIN, OpenAI says every locked chat, file, and piece of content tied to it is permanently wiped. You also can't lock temporary chats, archived chats, group chats, shared link sessions, or chats running scheduled tasks and meeting recordings.

The feature is rolling out across Free, Go, Plus, Pro, Business, Enterprise, and Edu plans, though ChatGPT for Healthcare is left out.

Device privacy, not cloud secrecy

OpenAI was unusually blunt about what this lock actually does: it keeps local peepers away from your screen, but it does not make you invisible to OpenAI.

Locking a thread does not enable end-to-end encryption. It doesn't put your data in an encrypted vault, alter how long OpenAI retains logs, change your data training opt-out settings, or purge what ChatGPT already remembered about you beforehand.

It is basically a digital curtain for your screen, not a cryptographic vault.

Why it matters

We think this is one of those small usability tweaks that actually changes how people use the app daily.

People treat ChatGPT like a therapist, a tax advisor, and a creative brainstorming partner all at once. Giving users an easy way to hide conversations without losing context makes handing your phone to a coworker or partner way less stressful.

That said, it creates a potential illusion of security. Because it lacks end-to-end encryption, it does nothing to protect high-stakes corporate secrets or sensitive whistleblowing from internal platform access or data retention rules.

Still, giving the locked vault its own isolated memory partition is a brilliant product decision. As chatbots handle more intimate life tasks, isolating contexts will be table stakes. Expect Google and Anthropic to copy this setup soon.