Open AI What Gmail’s AI Can See — And What It Can’t
Nov 22, 2025
The podcast dives into Google's recent clarification on how Gmail's AI interacts with user content. It discusses what the AI scans and what remains private, addressing common misunderstandings. Concerns about default opt-in settings for training data arise, alongside practical advice on opting out. The necessity of using real email content for AI improvement is explored, along with industry-wide transparency issues. The host shares personal insights, highlighting the balance between functionality and user privacy.
AI Snips
Chapters
Transcript
Episode notes
Gmail Data Powers Gemini Enhancements
- Google is using Gmail content and attachments to improve Gemini-powered features across email.
- This goes beyond on-the-fly suggestions and can feed model training unless users opt out.
Default Opt-In Raises Privacy Concerns
- Many users found these training settings enabled by default instead of explicitly opting in.
- Default opt-in for data use unsettles people even with Google's anonymization and security promises.
How To Fully Opt Out Of AI Training
- Do disable two separate settings to opt out: smart features in Gmail Chat & Meet and Google Workspace smart features.
- Make sure both toggles stay off and hit save to stop feeding your email data into AI training.