Google Splits Gemini Privacy Controls to Separate Chat History From AI Training
Users can now keep searchable Gemini conversations without allowing those chats to be used to train Google's generative models.
Overview
- Google began rolling out the change this week, replacing the single Keep Activity toggle with two separate controls: a History setting that governs how long chats are saved and a distinct toggle that controls whether chat content can be used to train generative AI models.
- When History is turned off, Google says new conversations will still be retained temporarily for about two months to respond to users, process feedback, and protect safety, and those retained chats may be reviewed by humans for quality and safety purposes.
- The separate training toggle lets users opt out of letting Google use text and uploaded attachments from future chats for model training while still leaving their past or saved conversations accessible if they choose to keep History on.
- Google is rolling the settings out globally over the coming weeks, will notify users by email, and will default both new toggles to off for accounts that already had Keep Activity disabled.
- The change reverses a long-standing design in Gemini that forced users to trade off keeping a conversation archive against contributing data to model training and follows recent updates aimed at making stored facts and review practices more transparent.