Google is updating Gemini privacy controls so users can keep their chat history without letting Google use that data to train generative AI models. Multiple outlets report that the Gemini activity settings will be split, separating controls for chat history from controls for AI model training.

According to the coverage, the change addresses a long-standing “package deal” where chat history and AI training permissions were previously linked. Several outlets also note that the update rolls out gradually over coming weeks. The overall tone is that Google is unbundling these settings to give users more independent control over whether their interactions can be used for training, while still allowing them to retain past chats.

Across reports, the key practical takeaway is the presence of two distinct toggles—one governing whether chat history is saved, and another governing whether AI training uses that data—rather than forcing one choice to affect both.