News · 2026-10-26

ChatGPT locked chats hide conversations, but PIN reset deletes them

OpenAI has begun rolling out locked chats in ChatGPT, allowing users to hide selected conversations from regular history and search behind a four-digit PIN or supported device authentication. Resetting a forgotten PIN permanently deletes the locked chats and content protected only by that PIN. Merely forgetting the number does not trigger deletion, and locking does not create end-to-end encryption.

Key facts

• OpenAI's release notes date the feature to October 20, 2026.
• A four-digit account PIN or supported device authentication controls access to locked chats.
• Confirming a forgotten-PIN reset permanently deletes protected content; OpenAI says Support cannot recover it.
• The primary source is OpenAI's locked-chat guide, alongside the release notes.

The release-note heading, “Introducing locked chats,” describes a feature with a fairly specific purpose. It removes a conversation from the ordinary places someone using the account might casually encounter it. That can be useful for a sensitive conversation, a personal project, or any material a user does not want visible in the regular chat list.

OpenAI says locked chats are hidden from normal history, pinned chats, and regular conversation search. Access is gated by the account's PIN or supported device authentication. Those changes concern the interface and the access step. They do not, by themselves, specify a different rule for every other copy or use of the underlying conversation.

A useful analogy is a locked drawer inside an office. The drawer changes who can casually open the material from that room. It does not automatically change the organization's filing policy, retention schedule, or other copies already held elsewhere. In the same way, hiding a conversation is a narrower protection than changing the service's storage and processing rules.

The recovery design is consequential. A person who forgets the PIN has not thereby instructed ChatGPT to erase anything. Deletion follows the reset flow: the user verifies identity and confirms resetting the forgotten PIN. OpenAI says that process permanently deletes all locked chats, information remembered only within them, and files or other content protected only by the previous PIN.

Changing the PIN while the current one is known is a different operation and does not delete the conversations. That distinction should appear in any account of the feature because the shorthand that forgetting a PIN erases chats implies an automatic trigger that the documentation does not describe. A user can forget a number without the service knowing that they have forgotten it.

The Help Center guide says Support cannot recover content deleted through the reset. Readers therefore need to understand the consequence before confirming that operation. The documentation establishes a product rule; it does not justify guessing why OpenAI chose that recovery architecture or what cryptographic design might sit underneath it.

OpenAI explicitly says locking does not create end-to-end encryption or a separate encrypted vault. It also does not change retention or the model-improvement setting, remove information already remembered outside the locked chat, or override applicable workspace policies. These boundaries matter because the same interface word can suggest much broader protection than the feature actually offers.

Memory is especially easy to confuse with conversation visibility. A detail may already have been retained separately before a chat was locked. Locking the conversation does not erase that existing memory. OpenAI's memory documentation describes separate controls for remembered information and reference to previous chats. Our agent-memory lesson explains why a retained summary can persist independently of its source conversation.

The rollout descriptions also have different scopes. The October 20 release notes name Free, Go, Plus, and Pro on the web and major mobile and desktop applications. The Help Center also lists Business, Enterprise, and Edu, while noting gradual availability. An account may therefore not expose the feature immediately, and a workplace account can remain subject to its organization's access and compliance requirements.

The supplied October 26 feed contains a post about the PIN feature, but no comments that establish how users are receiving it. There is no independent security audit in the dossier demonstrating additional guarantees beyond OpenAI's stated behavior. Those absences are reasons to keep the claim narrow, not to invent a security failure or a security endorsement.

The practical value is a new way to limit casual exposure inside the product. Its practical limit is that visibility, memory, retention, training preferences, and account recovery are separate controls. The clearest reading of the release is to use each control for the purpose OpenAI actually documents and to treat the reset confirmation as a permanent deletion decision.

Primary source, verified: read the paper →

Ground Truth. (2026, October 26). ChatGPT locked chats hide conversations, but PIN reset deletes them. Ground Truth. https://groundtruth.day/news/chatgpt-locked-chats-pin-reset-permanently-deletes-content.html

@misc{groundtruth:chatgpt-locked-chats-pin-reset-permanently-deletes-content,
  title  = {ChatGPT locked chats hide conversations, but PIN reset deletes them},
  author = {{Ground Truth}},
  year   = {2026},
  month  = {oct},
  url    = {https://groundtruth.day/news/chatgpt-locked-chats-pin-reset-permanently-deletes-content.html}
}

Topics: chatgpt · privacy · product-release · memory · data-controls