Are ChatGPT Locked Chats encrypted? What the PIN hides, and what it doesn't

The Morse teamPublished October 25, 20268 min read

ChatGPT's Locked Chats are not end-to-end encrypted. OpenAI's help center says locking a chat does not "create end-to-end encryption or a separate encrypted vault" (OpenAI). The lock is a 4-digit PIN, or Face ID, Touch ID, or device unlock on a phone, and it hides the chat in the app while OpenAI keeps it on its servers under the same retention, training, and legal rules as before.

Locked chats get the same encryption every ChatGPT chat gets. OpenAI says your content is "encrypted at rest and in transit between you and OpenAI" (OpenAI). That guards the chat against someone on your network or someone who steals a disk. OpenAI holds those keys, so its servers can decrypt the chat, and they have to for the model to answer you.

We read OpenAI's announcement, its help article on locked chats, its release notes, and its retention, memory, data controls, and privacy pages on October 20, 2026, the day the feature launched. OpenAI is rolling it out gradually, so some accounts don't have it yet.

What does ChatGPT's Locked Chats feature do?

It moves a chat out of your regular history into an area that opens with a PIN. OpenAI announced it on October 20, 2026 as a way to "keep sensitive conversations out of view," moving chats "out of your regular history and search into a protected space you can unlock with a PIN on web and desktop" (ChatGPT on X). The announcement doesn't mention encryption.

The help article gives the details (OpenAI):

• Where they go. Locked chats sit under Projects, then Locked chats, and drop out of your regular history, pinned chats, and conversation search.
• How they open. With a 4-digit account PIN, or on supported phones with Face ID, Touch ID, or your device unlock.
• When they close. On the web, refreshing or closing the tab ends access. On iOS and Android, access ends after 1 minute away from the chat or in the background. Lock now ends it at once.
• Who gets it. Free, Go, Plus, Pro, Business, Enterprise, and Edu, on the web, the iOS and Android apps, and the new desktop apps for macOS and Windows. Not ChatGPT for Healthcare.
• What can't be locked. Temporary, archived, and group chats, chats made from a shared link, and chats with scheduled tasks, among others.

OpenAI says the feature "can help keep sensitive conversations out of view when you share your screen or use a shared device."

Who does a locked chat protect you from?

People who can see your screen or pick up your device. Someone who borrows your laptop or watches your screen share sees a Locked chats project with the titles hidden, and needs your PIN to go further. On Android, screenshots of a locked chat come out blank.

Anyone who learns your PIN can open every locked chat, and OpenAI notes that anyone who can open one "can still copy its contents." Other devices where ChatGPT is already open "may continue to show the chat title until their conversation lists refresh," and on iOS, content that was already loaded may stay visible until the app refreshes.

What locking a chat doesn't change

OpenAI's help article lists what locking does not do. Each item is set below against what OpenAI's other pages say, read on October 20, 2026.

Can OpenAI see your locked chats?

As much as it can any other chat, because locking doesn't change who at OpenAI can get to one. Under the heading "Authorized human access to content," OpenAI's page on consumer data tells users: "Please do not enter sensitive information that you would not want reviewed or used." The same page says OpenAI may send "select portions of content to trusted service providers" for "data annotation and safety purposes" (OpenAI). Neither passage makes an exception for locked chats. Reporting on contractors who read ChatGPT conversations is covered, with what's alleged and what isn't proven, in our Project Lily explainer.

Can locked chats be subpoenaed?

Yes, as any stored chat can. A court order reaches what a company keeps, and a locked chat is kept like any other. OpenAI's privacy policy says it may hold data past your deletion when it is "legally required to retain your data (for instance, we receive a lawful subpoena)" (OpenAI). We cover how AI chats have reached court in can AI chats be subpoenaed, and the order that made OpenAI preserve deleted chats in the ChatGPT court order timeline.

What happens if you forget your ChatGPT PIN?

Forgetting it deletes nothing, but resetting it deletes every locked chat. OpenAI says a reset "permanently deletes all locked chats, remembered information stored only within locked chats, and files or other content protected only by your previous PIN," and that "OpenAI Support cannot recover the deleted content." A workspace admin can't reset it for you. If you know your current PIN, you can change it under Settings, Security and login, Account PIN without losing anything.

The same help article then says that after a reset, "Deleted conversations then follow the standard retention policy" (OpenAI). For ChatGPT, that means OpenAI schedules them "for permanent deletion from its systems within 30 days," unless a chat was already de-identified or OpenAI "must retain it longer for security or legal obligations" (OpenAI). The reset clears the chats from your account at once. OpenAI's copy goes on the usual schedule.

How to check your ChatGPT settings before you lock a chat

Locking leaves your other settings as they were, so check them yourself:

1. Training. Open Settings, then Data controls, and turn off Improve the model for everyone. Your new conversations, locked or not, then won't be used for training. A thumbs up or down still lets "the entire conversation associated with that feedback" be used (OpenAI).
2. Memory. Open Settings, Personalization, then Memory, and delete anything ChatGPT took from the chat before you locked it. Locking won't remove it.
3. Shared links. If you ever shared the chat, delete that link separately. Locking leaves it working.
4. Other devices. Refresh ChatGPT anywhere else it's open before you share that screen.
5. The PIN itself. Under Settings, Security and login, Account PIN, you can change it or turn on Face ID or Touch ID.

For a chat you don't need to keep, Temporary Chat does more on OpenAI's side: it stays out of history, doesn't create memories, isn't used for training while temporary, and "may be retained for up to 30 days for safety purposes" (OpenAI). Our guide to whether ChatGPT trains on your data has the rest of the training settings.

How is end-to-end encryption different?

With end-to-end encryption, your device encrypts a message so that only the other end can open it, and the service carrying it has no key. Locked Chats leaves the chat readable on OpenAI's side and puts a gate in front of it in the app. To test any service that calls itself encrypted, ask whether the company can open your message on its own servers. For ChatGPT, OpenAI's servers decrypt your message so its models can answer it.

An AI chat can be end-to-end encrypted only if the model runs somewhere you can check: a secure enclave whose hardware proof your device verifies before it encrypts anything. We explain how that works in end-to-end encrypted AI chat, and the hardware in what is a trusted execution environment.

How Morse keeps a sensitive chat off company servers

A PIN can only hide a chat that a company still holds. Locked Chats puts the lock on your screen, and the chat stays on OpenAI's servers, where retention, training settings, and legal requests all reach it. Morse keeps your chat history in your browser instead of on our servers, and in E2EE, our servers can't read a message on its way to the model either.

Morse gives each chat a privacy mode. Only E2EE is end-to-end encrypted:

• E2EE, for a chat no company should read. Your browser checks the enclave's cryptographic proof itself, then encrypts the message on your device for that enclave alone. Our servers relay text they can't read, and the reply is decrypted only in your browser. Open-weight models like GLM, Kimi, and DeepSeek answer inside the enclave, and the guarantee rests on the enclave hardware and attestation working as designed. Every E2EE reply has a lock you can open to see each check that passed.
• Private, for everyday chats no host should keep. Open-weight models, plus xAI's Grok, run on hosts that agreed to keep nothing: Morse has zero data retention agreements with its inference, search, and other providers. The host reads the message to answer it, without knowing it's you, so this mode isn't end-to-end encrypted.
• Anonymous, for GPT, Claude, and Gemini without your account. Our proxy removes your account, email, and IP address and sends the message under Morse's own account. This mode isn't end-to-end encrypted either: the provider sees the message without knowing it's you, and may keep it for a limited time to check for abuse, under its own policy.

In every mode, your chat history is stored in your browser, encrypted, and not on Morse's servers. It's locked when you sign out, and unlocking it takes two keys, one that never leaves your device and one our server gives you when you sign in, so a copy of your browser data alone can't be read. That storage encryption isn't end-to-end: as our privacy policy puts it, "we hold the key, but we never receive the data it unlocks." Morse doesn't train AI on your chats, sell your data, or show ads, so there's no training switch to check. Because we don't store chats, we usually have very little to give authorities. Clearing your browser deletes your chats for good, and we can't recover them, so save a copy of anything you need.

Which to use for what

• Someone using your phone or watching your screen: in ChatGPT, Locked Chats does this job. On Morse, your history is locked whenever you sign out.
• OpenAI's servers, training, and legal requests: a lock in the app doesn't help. On Morse, start the chat in E2EE, and only the verified enclave can read it.
• GPT without your name on it: Morse's Anonymous mode sends it under Morse's account, without your email or IP address.

E2EE chats come with Pro, at $19 a month, so you can start free, upgrade, and open the lock on your first reply today.