• Google is rolling out a feature called AI Inbox, which uses its Gemini AI (Gemini 3) to give you a more proactive view of your email — like a personalized briefing listing high-priority tasks and summaries instead of a plain chronological list. The AI will highlight suggested to-dos, key topics, and important messages.
• The overall Gmail overhaul also includes AI Overviews (thread summaries), Help Me Write, Suggested Replies, and enhanced proofreading tools.
• These features are initially being tested with a small user group in the United States and will roll out more broadly later.

Google’s stance on privacy and training

No email content used to train Gemini

• Google has clearly stated that your personal emails and attachments will not be used to train its foundational AI models (like Gemini). Processing is done in a secure, isolated environment that keeps your inbox content within that private boundary.
• You still retain control: all AI features can be switched off if you don’t want Gmail using your email content for AI-generated summaries or suggestions.

What about broader privacy concerns?

Some tech commentators and experts have raised concerns that Gmail’s “Smart features” settings (which enable AI-powered tools) are opt-in and may allow Gmail to process email content — but this does not mean Google is feeding those emails into its AI model training data. Google distinguishes between using content to power features and using it to improve/fine-tune models, and claims only the former happens for these tools.

Key takeaway

• Yes, Gmail’s AI Inbox and other tools use Gemini to analyze your email content to generate summaries, to-dos, and smarter insights.
• No, Google says it will not use your emails to train its AI models — that data isn’t added to Gemini’s underlying training data sets.
• You can opt out of these AI features if you prefer not to have your email processed for any AI-generated suggestions.