Mechanism: The feature relocates selected conversations from ordinary views and gates access through an account PIN or supported device authentication, while leaving encryption, retention, and model-improvement settings unchanged.
Why now: OpenAI says Locked Chats are rolling out gradually across Free, Go, Plus, Pro, Business, Enterprise, and Edu plans, so users are encountering a new privacy control whose label could be mistaken for stronger protection.
First implication: Treat the feature as a way to reduce casual visibility during shared-device or screen-sharing use, not as a protected vault for sensitive material.
Evidence boundary: The official help page documents feature behavior and stated limitations; it does not establish the rollout completion date or independently measure how much exposure the feature prevents.
The feature changes where a conversation appears and how it is opened; the source does not claim it changes underlying retention, encryption, or training behavior. That boundary is the operational point: interface privacy can help with exposure in ordinary use, while separate policies and controls remain necessary for confidentiality and data governance.
First watch condition: Watch whether OpenAI adds a separate documented encryption or retention control, rather than inferring one from the Locked Chats label.
For operators, PIN recovery also has a consequential failure mode: resetting a forgotten PIN deletes the locked material rather than restoring access. This makes account recovery and the availability of device authentication worth considering before relying on the feature for important conversations. The public documentation does not establish the rollout's completion date or independently measure how much exposure it prevents.