← tutti gli articoli

ChatGPT can now lock a chat behind a PIN. Here is what it does and what it doesn't

OpenAI's new Locked Chats hides chosen conversations behind a 4-digit PIN or Face ID, even on the Free plan. What it does, a real example, how to try it in three steps, and the limits worth knowing.

·4 min di lettura

Most of us have done it. You hand your phone to a friend to show them one ChatGPT answer, and your chat list is right there: the medical question, the salary draft, the surprise party plan. On 20 October, OpenAI started rolling out a small feature for exactly that moment: Locked Chats.

I work with data all day, and the first thing I ask about any "privacy" feature is simple: what does it protect, and from whom? So here is the short version, with the limits included.

What it does

You choose a conversation and lock it. The chat leaves your normal history and moves to a separate Locked Chats section. It no longer shows up in your chat list, in pinned chats or in search.

To open the section you need a 4-digit PIN for your account. On supported phones you can also use Face ID or Touch ID. According to OpenAI's help article, access closes again on its own: on the web when you refresh or close the tab, and in the mobile apps after one minute away from locked content.

There is one detail I like a lot. If you use Memory, locked chats get their own separate memory. New messages in a locked chat can update that memory, but they don't update your regular Memory. So what you say there doesn't show up later in your everyday chats.

This is also different from Temporary Chats. A temporary chat disappears from your history. A locked chat stays, and you can come back to it whenever you want.

A concrete example

Say you are planning a career change. You use ChatGPT to rewrite your CV, compare two job offers and prepare for an interview. You still use the same phone or laptop with your partner or your kids, and sometimes you share your screen in a meeting at work.

You lock those three chats. Your regular list shows only your recipes and spreadsheet questions. When you need the job chats, you open Locked Chats, type your PIN, and continue where you left off. The CV advice is still there, and the memory from those chats stays in that section.

Try it in three steps

  1. Open ChatGPT on the web or in the app and update it if needed. The feature is rolling out gradually, so you may not see it yet.
  2. Open a conversation you want to keep private and look for the lock option for that chat. The first time, ChatGPT asks you to set a 4-digit PIN.
  3. Find your chat in the Locked Chats section and open it with your PIN, or with Face ID or Touch ID on your phone.

OpenAI lists the Free, Go, Plus, Pro, Business, Enterprise and Edu plans. It is not available in ChatGPT for Healthcare.

What it does NOT do

This part matters more than the feature itself.

  • It is not encryption. OpenAI says locking a chat does not create end-to-end encryption or a separate encrypted vault. It protects the chat from people who use your device, not from anything else.
  • Nothing changes on OpenAI's side. It does not change how long the chat is kept or your "Improve the model for everyone" setting. It also doesn't remove what ChatGPT already remembered before you locked the chat.
  • Forgetting the PIN is expensive. If you reset it, all locked chats, and the files and memory protected only by the old PIN, are deleted for good. OpenAI says its support team cannot get them back.
  • Work accounts are different. On Enterprise and Edu, the PIN does not stop authorized access through OpenAI's Compliance Platform, so your organisation can still reach those messages.
  • Not every chat can be locked. TechCrunch lists temporary, archived and group chats, chats from a shared link, chats with scheduled tasks and chats with meeting recordings.

Also this week: Claude Duet

The AI launch with the most buzz on LinkedIn this week was a different one. On 21 October, Anthropic opened a research preview of Claude Duet, an app where Claude works in the background. It sends a morning brief, checks your connected apps every few hours, and asks for approval before it sends a message, makes a payment, signs a document or deletes anything. Claude's LinkedIn post about it had more than 3,700 reactions within a day. For now it is waitlist only, for Pro and Max subscribers in the US, so most of us can't try it yet.

My take

Locked Chats doesn't make ChatGPT more private. It makes your screen more private, and that is a real, everyday problem. I would use it, set a PIN I won't forget, and keep the most sensitive things (health records, passwords, ID numbers) out of any chatbot, locked or not.

Sources

aichatgptprivacy