ChatGPT Locked Chats Explained: The Essential Privacy Guide





ChatGPT Locked Chats Explained: The Essential Privacy Guide


On October 20, 2026, OpenAI quietly shipped one of the most requested privacy features in ChatGPT’s history: ChatGPT Locked Chats. For the first time, you can take any individual conversation, pull it out of your regular chat history, and seal it behind a 4-digit PIN — or behind your phone’s Face ID, Touch ID, or device unlock. Anyone who picks up your phone, glances at your laptop, or watches your screen during a presentation no longer gets a free look at your most sensitive threads.

But there is a catch that matters more than the feature itself, and OpenAI has been unusually blunt about it: ChatGPT Locked Chats hide your conversations; they do not encrypt them. Locking a chat changes nothing about how long OpenAI keeps it, whether it can be used to train models, or what ChatGPT already remembers about you. And there is one sharp edge you must know before you lock anything — if you ever reset a forgotten PIN, every locked chat is permanently deleted, with no recovery possible. Not even OpenAI’s support team can bring them back.

In this guide, I break down everything we know about ChatGPT Locked Chats — how to set them up on the web, desktop, iPhone, and Android, how the separate memory system works, exactly when the lock re-engages, the full list of things the lock does not do, and which chats cannot be locked at all. I have pulled together the launch reporting from multiple outlets and OpenAI’s own help documentation so you get the facts straight, with honest caveats where they matter.

Table of Contents

What Are ChatGPT Locked Chats?

ChatGPT Locked Chats is a per-conversation privacy feature that moves selected chats into a protected space you can only open with a 4-digit account PIN or a supported biometric unlock. It is the first time any major AI chatbot has offered this kind of protection — the ability to hide some conversations while leaving the rest of your history untouched.

ChatGPT Locked Chats: private AI conversation secured on a smartphone at night

The idea is simple. Millions of people now use ChatGPT for deeply personal things — relationship advice, health questions, financial planning, work frustrations they would never put in an email. Until now, every one of those threads sat in the same sidebar as everything else, visible to anyone who could see your screen. OpenAI’s own examples for the feature, reported by TechCrunch, are telling: personal conversations and planning a surprise. The mental model is “hand your phone to a friend without sweating,” not “hide from your employer or OpenAI.”

The feature arrived with a post from the @ChatGPT account on the morning of October 20, 2026, a release-notes entry titled “Introducing locked chats,” and a dedicated help article. There was no big OpenAI blog post and no executive commentary — the company treated it as a straightforward product update, not a marketing moment. Notably, Android Authority had already spotted the feature hiding in the app’s code back on August 21, two months before launch, so the rollout itself was not a surprise to close watchers.

Where ChatGPT Locked Chats Live: Projects > Locked Chats

When you lock a conversation, it leaves your normal chat history, your pinned chats, and the search bar. It moves to its own dedicated section that OpenAI’s help documentation calls Projects > Locked Chats — a separate area of the app reserved for protected threads.

To open anything inside, you enter your PIN or use your device’s biometric unlock. The same 4-digit PIN works on every device signed in to your OpenAI account, and on phones the PIN always remains available as a fallback behind the biometric option. So if Face ID fails in bad lighting or your fingerprint sensor has a bad day, you can still type the PIN to get in.

One small but important detail: unlocking is per-session, not permanent. As soon as you leave the locked area — or one of the auto-relock triggers fires, which I cover below — the gate closes again and the PIN or biometric check returns. There is no “stay unlocked for the day” setting.

How to Lock a ChatGPT Chat: Step-by-Step Setup Guide

Locking a chat takes about ten seconds. Open the conversation you want to protect, open its options menu (the three-dot menu on the chat), and choose Lock chat. Confirm, and you will be asked to create your 4-digit account PIN the first time. That is it — the chat immediately moves to the Locked Chats area.

Setting up ChatGPT Locked Chats: creating a PIN to lock a conversation on a laptop

On the web and desktop apps

On chat.openai.com and in the new ChatGPT desktop apps for macOS and Windows, the lock is PIN-only. Open the chat’s three-dot menu, choose Lock chat, confirm, and set your 4-digit account PIN if you have not already. From then on, opening anything in the Locked Chats section asks for the PIN. Your locked chats re-lock when you leave the tab or window, refresh the page, or close the tab — so the gate closes behind you automatically.

On iPhone and Android

On phones and tablets, the protection goes a step further. After locking a chat, you can open it with your PIN, or switch on the biometric option: Face ID or Touch ID on iPhone, and your device’s own unlock method on Android (listed in the app as “Use device unlock”). The flow is the same — lock from the conversation’s menu — but unlocking is one glance or one tap instead of four digits. As reported across the launch coverage, the PIN always stays as a fallback and is identical on every device tied to your account.

Access on mobile expires on its own: after one minute away from a locked chat, or after one minute with the app in the background, the lock re-engages. A Lock now button ends access immediately anywhere, and signing out or switching accounts closes the gate everywhere too.

What if the “Lock chat” option does not show up?

The feature is rolling out in stages, so the option may not be on your account yet. If it is missing, update the ChatGPT app to the latest version first — several outlets confirmed that is the most common reason the option is absent. If it still does not appear, your account simply has not been switched on yet; OpenAI is enabling it gradually across all tiers, and there is no manual opt-in to force it.

Unlocking and Re-locking: When Access Expires

The re-lock behavior of ChatGPT Locked Chats is stricter than most people expect, and honestly that is the point. Here is the full set of rules, from OpenAI’s help documentation:

  • Web and desktop: locked chats re-lock when you leave the tab or window, refresh the page, or close the tab.
  • iPhone and Android: access ends after one minute away from a locked chat, or after one minute with the app in the background.
  • Everywhere: the “Lock now” button ends access immediately, and signing out or switching accounts does the same.

ChatGPT Locked Chats during a work meeting: keeping sensitive conversations hidden on a shared screen

Think of it this way: the moment you stop looking at a locked chat, the clock starts on re-locking it. On the web it is effectively instant — a tab refresh is enough. On mobile you get about sixty seconds of grace. This design is clearly built for the “shared screen” scenarios OpenAI keeps naming: a colleague glances at your laptop, a family member picks up your phone, you share your screen in a meeting. In each of those cases, the lock closes before a wandering eye has time to work.

ChatGPT Locked Chats and Memory: The Two-Way Rule

This is the part of ChatGPT Locked Chats that deserves the most careful reading, because memory behaves asymmetrically — and that asymmetry is deliberate.

Here is what happens, in plain language:

  • New messages in locked chats do NOT update the memory your regular chats use. Anything you type inside a locked chat stays out of the memory pool that your normal conversations draw on.
  • Locked chats keep their OWN separate memory that does not transfer to regular conversations. Your locked threads build a private context that stays inside the locked area.
  • Locked chats can still READ your regular memory. Per OpenAI’s help article, the partition is one-directional: locked chats draw on the context your regular chats have already stored, but they never write back to it. So a locked chat may reference things you told ChatGPT in ordinary conversations — the door only swings one way.

Why does this matter? Imagine you use ChatGPT to plan a surprise party in a locked chat. With the two-way rule, the party details never leak into your everyday chats — your partner will never see a stray memory reference to the surprise in an unlocked thread. But the locked chat itself can still know things like your name, your city, or your preferences from your regular memory, which makes it more useful than a completely amnesiac session.

ChatGPT Locked Chats memory rules explained: how separate memory works in locked conversations

There is one more thing the lock does not erase: anything ChatGPT remembered before you locked the chat. If a thread spent three weeks in your regular history building up memory entries, locking it today does not retroactively pull those entries out of your memory. The partition starts at the moment of locking, not before. If you want old memory gone, you have to delete those memory entries yourself, separately.

The PIN Warning: Why Resetting Deletes Everything

This is the single most important paragraph in this entire guide, so read it twice.

If you forget your PIN, the only recovery option OpenAI offers is a PIN reset — and resetting the PIN permanently deletes every locked chat and every file protected by it. There is no recovery flow, no cooling-off period, no way back. OpenAI’s support team cannot restore them. A workspace administrator cannot reset anyone’s PIN for them. Gone means gone.

This is not a bug or an oversight; it is the security design. The PIN is the only key to the locked area, and OpenAI is not keeping a spare. The upside is that nobody — not a hacker who compromises your account credentials, not an employer, not OpenAI itself — can bypass the gate to read your locked chats. The downside is that the protection is absolute in the wrong direction too.

The practical rules, then, are simple:

  • Do not rely on memory alone. Write the PIN down somewhere safe, or lean on biometrics on your phone so the PIN is only a backup.
  • Changing your PIN is safe as long as you still know the current one — a normal PIN change deletes nothing. Only a reset (used when the PIN is forgotten) destroys everything.
  • Keep a copy of anything irreplaceable that lives only in locked chats — long personal threads, uploaded documents, created files. Files uploaded or created inside locked chats stay hidden with them, and they go down with a PIN reset too.
  • One PIN, every device. The same PIN works everywhere you are signed in, so a reset on your phone wipes the locked area on your laptop as well. Think of the locked space as one vault with one key — and if you would rather agents log in for you without ever seeing your passwords, see our 1Password for Agents explainer.

What ChatGPT Locked Chats Do NOT Do

OpenAI’s help page includes an unusually long list of things the lock does not change. Here it is, because this is where most people’s assumptions go wrong:

  • No end-to-end encryption. Locking a chat does not encrypt it. OpenAI’s servers store locked chats exactly as before.
  • No separate encrypted vault. There is no special hardened storage — the lock is an access gate on the same data, not a new safe.
  • No change to data retention. Locked chats are kept for the same length of time, under the same policies, as unlocked ones.
  • No change to model training settings. The “Improve the model for everyone” toggle applies exactly as it did before. If you want locked conversations excluded from training, you have to turn that setting off separately — the lock does not do it for you.
  • No memory wipe. As covered above, locking does not remove what ChatGPT already remembered from the conversation’s unlocked days.
  • No invisibility from compliance tools. On Business, Enterprise, and Edu plans, employer or school compliance and e-discovery tooling sees the same data it always did. A lock is not a shield against your organization.

The honest summary: ChatGPT Locked Chats protect you from people looking at your screen, not from OpenAI’s infrastructure, not from model training, and not from institutional oversight. Notebookcheck’s coverage put it neatly — the feature is “locked, not encrypted” — and that distinction should be the first thing anyone understands before relying on ChatGPT Locked Chats.

The Chats You Cannot Lock

Not every conversation is eligible for ChatGPT Locked Chats. According to the launch reporting and OpenAI’s documentation, you cannot lock:

  • Temporary chats — the sessions that already stay out of your history. They cannot be saved as locked chats either.
  • Archived conversations — move a chat out of the archive first if you want to lock it.
  • Group chats — shared threads with other people cannot be locked by one participant.
  • Shared-link chats — anything you have generated a share link for.
  • Chats with scheduled tasks — threads that have automations attached.
  • Chats containing meeting recordings — the new meetings capture feature is excluded.

The pattern is sensible: anything with a second audience (shared links, groups), a second system reading it (scheduled tasks), or an already-private mode (temporary chats) is out of scope. The lock is designed for ordinary one-on-one conversations you want to keep one-on-one.

ChatGPT Locked Chats vs Temporary Chats

ChatGPT already had a privacy feature before this one — Temporary Chat — and the two are easy to confuse. They solve opposite problems:

  • Temporary Chats open a separate session that never enters your history at all, and (while temporary) its contents are not used to train models. But it is use-it-and-lose-it: you cannot revisit a temporary chat later.
  • ChatGPT Locked Chats stay saved and revisit-able at any time — they follow your account’s normal training setting — but they are hidden behind the PIN gate and their new messages do not feed your regular memory.

Pick Temporary Chat when you want a conversation that leaves the smallest possible trace and you do not need to return to it. Pick ChatGPT Locked Chats when you want to keep a thread long-term — a months-long personal project, an ongoing private journal — but want it out of casual view. They can coexist: use temporary chats for one-off sensitive questions and locked chats for sensitive threads you intend to continue.

Worth noting: the lock cannot be applied to a temporary chat, so you have to choose at the start which kind of privacy you want for a given thread.

Who Gets Locked Chats: Plans, Devices, and the Rollout

ChatGPT Locked Chats is rolling out across every consumer and business tier: Free, Go, Plus, Pro, Business, Enterprise, and Edu. That includes the free plan — this is not a paywalled perk. The one exception is ChatGPT for Healthcare, which is excluded (likely because healthcare accounts run under stricter data-handling rules where a client-side gate could complicate compliance).

Device coverage is broad: the web, the new ChatGPT desktop apps for macOS and Windows, and the iOS and Android apps all support it. The PIN is the universal key; biometrics are the mobile convenience layer. Because this is a phased rollout, availability varies by account — if the “Lock chat” option has not appeared for you yet, updating the app is the first step, and waiting is the second.

For US users, the important pricing note is that there is no price at all: no plan upgrade, no add-on, no per-chat fee. Free-tier users get the same lock as Pro users. Given that OpenAI is aggressively monetizing advanced features elsewhere, shipping this one free to everyone is a notable — and welcome — choice.

ChatGPT Locked Chats on mobile: unlocking private conversations with Face ID and fingerprint

When Should You Use ChatGPT Locked Chats?

The feature is at its best in everyday, human situations — not spy-movie ones. Here are the scenarios where locking a chat genuinely earns its keep:

  • Shared devices. A family laptop, a phone a kid borrows for games, a tablet on the coffee table — anywhere someone else’s fingers might wander into your app.
  • Screen sharing at work. Presenting in a meeting when a stray notification or an accidentally opened sidebar could flash a personal thread to the whole call. Locked chats simply are not in the sidebar.
  • Relationship and health conversations. The threads people most want private — dating advice, therapy-adjacent journaling, medical questions — are exactly the ones that feel wrong sitting in an open list.
  • Surprises and gifts. OpenAI’s own example: planning a surprise party or a gift in a thread your partner might glimpse on a shared device.
  • Financial planning. Budget conversations, debt paydown plans, salary negotiations — private, ongoing threads you revisit for months. This pairs naturally with the site’s Claude Duet coverage of AI assistants that touch your money: the more an assistant knows about your finances, the more sense it makes to gate the thread.
  • Journaling and personal projects. A long-running private journal or creative project that you want to keep — unlike a temporary chat — but never want surfacing in search.

And the scenarios where it does not help: anything where the threat is OpenAI’s data handling, your employer’s compliance tooling, or a legal discovery process. For those, the answer is still the same as before — do not put it in ChatGPT at all. The lock is a curtain, not a safe. This is the same hard lesson from the recent Meta Muse privacy leak story we covered: convenience features around AI assistants are not the same thing as data protection.

Frequently Asked Questions

Are ChatGPT Locked Chats encrypted?

No. ChatGPT Locked Chats hide conversations behind a PIN or biometric gate, but OpenAI does not apply end-to-end encryption, does not move the data to a separate encrypted vault, and does not change retention or model-training policies. Think of it as a locked drawer, not a safe-deposit box.

What happens if I forget my ChatGPT Locked Chats PIN?

The only option is a PIN reset — and resetting the PIN permanently deletes every locked chat and every file in the locked area, with no recovery. OpenAI’s support cannot restore them, and a workspace admin cannot reset a user’s PIN. Changing the PIN while you still know the old one, by contrast, deletes nothing.

Do ChatGPT Locked Chats stop OpenAI from using my conversations for training?

No. The “Improve the model for everyone” setting applies to ChatGPT Locked Chats exactly as before. To keep locked conversations out of training, turn that setting off separately in your ChatGPT data controls.

Do messages in ChatGPT Locked Chats affect ChatGPT’s memory of me?

New messages in ChatGPT Locked Chats do not update the memory your regular chats use, and locked chats keep their own separate memory that does not transfer out. But they can still read your existing regular memory — the partition is one-directional. And locking does not erase anything ChatGPT remembered from the chat before it was locked.

Is ChatGPT Locked Chats available on the free plan?

Yes. The rollout covers Free, Go, Plus, Pro, Business, Enterprise, and Edu at no extra cost. The only excluded tier is ChatGPT for Healthcare.

Why don’t I see the “Lock chat” option yet?

It is a phased rollout. Update the ChatGPT app to the latest version first — that is the most common fix. If the option still does not appear, your account has not been enabled yet, and there is no manual way to force it.

Can I lock a chat on my phone and open it on my laptop?

Yes. The same 4-digit PIN works on every device signed in to your account. On the web and desktop you enter the PIN; on mobile you can use Face ID, Touch ID, or device unlock, with the PIN as fallback.

Can I unlock a chat and move it back to normal history?

OpenAI’s documentation describes locking as a move into the Protected space; the widely reported flow is one-directional into the Locked Chats area. If you need a thread back in regular history, the safe approach is to copy the content you need into a new regular chat rather than assuming an “unlock” button exists on your version of the app.

How is this different from an incognito or temporary chat?

Temporary chats never enter your history and are not used for training while temporary, but you cannot revisit them. ChatGPT Locked Chats stay saved and revisit-able behind the gate, follow your normal training setting, and keep their own separate memory. Choose temporary for one-off questions you want to leave no trace of; choose locked for private threads you want to keep.

Final Verdict

ChatGPT Locked Chats is a genuinely useful feature that does exactly one thing well: it keeps your private conversations out of casual view on devices and screens you share. For the scenarios OpenAI names — a friend borrowing your phone, a family member on the family laptop, a screen-share at work — it is a clean, free, well-designed answer, and shipping it to every plan including Free is the right call.

But go in with clear eyes about ChatGPT Locked Chats. The lock is a curtain, not a safe: no encryption, no change to retention or training, no erasure of past memory, no shield from employer compliance tools. And the PIN reset rule is unforgiving — one forgotten PIN, and every locked thread is gone forever. Treat the PIN like a house key: keep a spare copy somewhere safe, or let your phone’s biometrics carry the load.

Used with those limits understood, Locked Chats earn a permanent place in the ChatGPT privacy toolkit — alongside temporary chats for the truly ephemeral stuff, and the training opt-out for everything you would rather OpenAI never learn. Privacy in AI assistants is a stack of small, honest tools, not one magic switch. This one just happens to be the most practical of the bunch.


Leave a Comment

Your email address will not be published. Required fields are marked *