TLDRocket
Sign in

ChatGPT now lets you lock private chats with a PIN—but forgetting it could erase them for good

Fortune Fernanda Tronco ● Covered by 2 sources

ChatGPT now has locked chats hidden behind a PIN. Forget it, and OpenAI says those chats are gone for good.

Based on reporting by Fortune, Fernanda Tronco — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

OpenAI has started rolling out a way for ChatGPT users to hide sensitive conversations behind a four-digit PIN. Locked chats stay in the account, but they disappear from the usual chat list, pinned chats, and search, and live in a separate spot under Projects > Locked chats.

The catch is brutal. If someone forgets the PIN and resets it, OpenAI says every locked chat gets deleted permanently, along with memories stored only there and any files or other content protected only by that PIN. The company says support can’t bring anything back, and workspace admins can’t reset it for users either. If the PIN is still remembered, changing it does not wipe anything.

This is less of a privacy vault than a visibility switch. OpenAI says locked chats do not add end-to-end encryption, do not create a separate encrypted store, and do not change how long the company retains the data. They also do not alter a user’s model-improvement setting, override workplace rules, or remove anything ChatGPT already remembered before the chat was locked. In Enterprise and Edu workspaces, the chats still show up in compliance logs.

The feature is rolling out across Free, Go, Plus, Pro, Business, Enterprise, and Edu plans. It works on the new macOS and Windows desktop app, and on mobile devices users can unlock with Face ID, Touch ID, or the phone’s own unlock method where supported. Voice works in locked chats on the web and in iOS and Android apps where available, but not on desktop.

Not everything can be locked. Temporary chats, archived chats, group chats, chats from shared links, scheduled-task chats, meeting recordings, Codex conversations, and local Work conversations are excluded, and the feature is not available on ChatGPT for Healthcare. Shared links are another weak point: locking a chat does not kill an existing link, so anyone who already has it can still view the old content. To revoke that access, users have to delete the link separately.

OpenAI is trying to make chats feel safer without actually changing the company’s underlying access to them, and that’s the real story here. It is privacy theater with a PIN, which is better than nothing and still miles short of what people usually imagine when they hear “locked.”

My take — AI-written commentary, not fact-checked reporting

This is the modern AI playbook in one neat little package: make privacy look simpler, then bury the actual tradeoffs in settings pages and footnotes. A PIN is not encryption, no matter how politely it is labeled. The industry keeps asking for trust while building products that only deserve careful, suspicious use.

Read more about this at: Fortune

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.