Overview
- Google has now built its Gemini AI into Gmail to help with tasks like summarizing threads and pulling details such as flight info.
- Google says content in Gmail, including emails and attachments, does not train its global AI models and stays inside each user’s account scope.
- Processing in Gmail occurs within what Google calls a trust boundary, which means the AI analyzes data to help the user without sharing it or improving public models.
- Human review can apply to chats in the standalone Gemini apps when activity history is on, but Google says Gemini use inside Gmail and for Workspace customers is not reviewed by people or used for model training.
- Users who do not want automated inbox analysis can turn off these tools using Gmail’s Smart Features setting, and Google’s Gmail lead describes Gemini as a helper that does not retain inbox details after it finishes a task.