A Lock on ChatGPT Changes the Economics of Private Conversation

ChatGPT has added the ability to lock individual chats, giving users a direct means of concealing sensitive exchanges when a borrowed phone, shared computer or family device is nearby. The company presents the feature as protection for personal discussions, surprise plans and other conversations users may not intend to expose. The practical change is modest in code but material in behaviour: privacy no longer depends solely on the user remembering to delete a conversation or on whoever holds the device choosing not to open it.

The feature arrives amid a persistent institutional tension between convenience and control. ChatGPT encourages sustained interaction and broad use, while many of the conversations people value most can contain intimate details, professional plans or information that would cause harm if disclosed. A lock can reduce the chance of accidental exposure, yet it cannot establish who holds the account, whether access credentials have been copied or how the underlying data is governed. The added control therefore transfers part of the security burden from the platform to the user at the precise moment when the user may assume the platform has taken responsibility.

The clear winner is OpenAI, which can present a visible answer to a familiar complaint without abandoning the wider model for conversation. Users gain a useful safeguard, but they also bear the risk of misplaced confidence in a single lock. Other services may follow with similar mechanisms, normalising the idea that private chats require an active act of protection. The larger outcome will be judged by what happens when locked conversations are lost, compromised or improperly accessed. Privacy in artificial intelligence will no longer be a promise made by the interface alone; it will be a relationship built through controls that users understand and systems that can defend.