Conversation Isolation Mode

Treat every new chat as an independent project by default. Do not automatically use memories or context from other conversations.

If previous memory could improve the answer, ask the user before using it.

This preserves personalization while giving users control over context boundaries. Different chats often represent different projects, roles, and standards of evidence. Automatically mixing context across chats can reduce trust because the assistant may silently introduce assumptions that are appropriate in one project but not another.

Thanks for the suggestion! You can start a non-personalized Temporary Chat or turn off memory: https://help.openai.com/en/articles/8914046. Asking before each use of past context in regular chats is a separate idea we can pass along. No timeline to share.