ChatGPT now lets you lock your chats

You can now password-protect sensitive conversations within ChatGPT, keeping them separate from your main chat history.

OpenAI has introduced a new "Locked Chats" feature for ChatGPT, allowing users to secure specific conversations with a four-digit PIN or biometric authentication (Face ID/Touch ID). This feature is designed to prevent unauthorized access to sensitive discussions when a device is shared. Locked chats are moved to a separate section, maintaining their own memory distinct from regular chat history and preventing data leakage into other conversations. While this adds a layer of privacy, OpenAI clarifies that it does not implement end-to-end encryption, alter data retention policies, or remove previously learned information. The feature is rolling out to all ChatGPT plans, excluding ChatGPT for Healthcare, and resetting the PIN will permanently delete all locked content.

The introduction of Locked Chats signifies OpenAI's response to growing user concerns about privacy, especially as AI models become integrated into more personal and sensitive use cases. By offering a dedicated, PIN-protected space, OpenAI aims to build trust and encourage broader adoption of ChatGPT for tasks involving personal information or confidential discussions. This move acknowledges the need for granular privacy controls within conversational AI, moving beyond simple deletion or temporary chat options to provide a more robust solution for immediate privacy needs.

This feature directly impacts users who engage in personal planning, share sensitive information, or simply wish to maintain a degree of privacy from others who might access their devices. The separation of memory between locked and regular chats is a key technical differentiator, ensuring that private conversations do not inadvertently influence AI responses in public or general chats. It suggests a market trend towards offering more tailored privacy settings for AI interactions, potentially influencing competitors to implement similar user-centric security measures.