- Users can keep chat history while opting out of model training
- Retention periods are adjustable with manual review and deletion
- Chats persist about two months after disabling history for responses and safety
AI CURATED HOT STORY · 11/18/2026
Google Decouples Gemini Chat History from Model Training
Google is rolling out a new privacy control for its Gemini app that separates chat history from model training consent into two independent toggles. Previously, users who wanted to opt out of training had to disable chat history, losing the ability to reference past conversations. Now, they can keep their chat history while opting out of training, gaining more granular control over their data.
According to multiple sources, the new setting allows users to adjust the retention period for chat history and manually view or delete records. Notably, training authorization remains opt-out by default, requiring users to actively disable it in settings. Additionally, when history is turned off, chat content is still retained for about two months for providing responses, handling feedback, and safety protection.
This change addresses users' dual needs for privacy and convenience, marking a key step for Google in the generative AI privacy race. It may prompt the industry to re-examine data usage transparency and encourage other vendors to follow with similar decoupled designs, seeking a better balance between compliance and user experience.
Hourly heat curve
9 fully observed hoursThe line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.
Coverage timeline
Total 2 reports · Latest first- IT之家 · 智能时代T2·38 pts
- Ars Technica AIT2·58 pts
- Gemini app adds a standalone model-training toggle separate from chat history
- Previously opting out of training required disabling history, making past chats unrecoverable
- Training consent remains opt-out by default; users must actively disable it in settings