Google has spoken out after widespread reports claimed Gmail was now feeding users’ emails and attachments into its Gemini AI training pipeline. The rumours sparked concern across social media, with many users believing that Google had silently changed its data policy — and that the only way to opt out was by disabling Gmail’s “smart features”.
However, Google has firmly denied these claims, calling the reports “misleading” and confirming that Gmail content is not used to train Gemini AI models.
Google Says Gmail Data Is Not Used for AI Training
In a statement to The Verge, Google spokesperson Jenny Thomson clarified that the company has not changed any user settings, nor is it using Gmail content to train Gemini:
“These reports are misleading – we have not changed anyone’s settings… and we do not use your Gmail content for training our Gemini AI model.”
According to Google, the confusion began after updated wording and repositioned settings for Gmail’s smart features appeared in some user accounts. These features — such as categorising emails, generating quick replies, spam detection and adding travel bookings to Calendar — rely on analysing message content.
But Google stresses that this analysis is strictly for powering Gmail functions, not for training AI models.
Why Users Thought Gmail Was Training AI
The misunderstanding escalated after a Malwarebytes article suggested Gmail content might be used for AI training and that users were being automatically opted in. That article has since been revised, admitting it “contributed to a perfect storm of misunderstanding.”
Malwarebytes now confirms:
-
Gmail’s smart features settings are not new
-
Email scanning powers only in-app features
-
This processing is separate from Gemini AI training
-
Google documentation indicates the settings are opt-in
Adding to the confusion, The Verge noted that at least one staff member observed their smart features being toggled back on unexpectedly, hinting at inconsistent rollout behaviour.
What Actually Changed in Gmail
In January, Google introduced more granular controls, splitting the old smart features toggle into two separate permissions:
-
Personalisation inside Google Workspace (Gmail, Calendar, Meet, etc.)
-
Personalisation across other Google services (Maps, Wallet, etc.)
Google says this update did not alter data handling practices but was designed to give users clearer control over how their data influences product experiences.
These settings may personalise:
-
Automatic addition of flight details from Gmail to Calendar
-
Surfacing Drive files in Gmail via Gemini
-
Pulling tickets or loyalty cards into Google Wallet
Importantly, none of these permissions allow Google to train Gemini on Gmail data.
Gmail for Android Gets Smarter Attachment Notifications
While clarifying data usage, Google is also rolling out a visual update to Gmail notifications on Android.
According to 9to5Google, users will now see:
-
A paperclip icon at the start of the subject line
-
A thumbnail preview replacing the profile photo when a photo attachment is present
-
For multiple images: a row of thumbnail previews
-
For documents: a labelled pill-style indicator showing part of the file name
Expanded notifications now prioritise the attachment preview over the email text, making it easier for users to spot important files at a glance.
Google Introduces Nano Banana Pro, Powered by Gemini 3 Pro Image
In related AI news, Google has launched Nano Banana Pro, the upgraded version of its Nano Banana image generator, now using the Gemini 3 Pro Image model.
The new version delivers:
-
Clearer, sharper visuals
-
Better multilingual text rendering
-
Richer graphics for diagrams and infographics
-
Higher consistency across images
-
Support for up to 14 blended inputs
-
Advanced creative controls such as lighting tweaks, region edits and high-resolution outputs
Nano Banana Pro is rolling out globally with limited free usage and expanded access for Plus, Pro and Ultra subscribers. It will also integrate with Workspace apps, Search AI Mode (US) and NotebookLM.
Conclusion
Despite the wave of online speculation, Google maintains that Gmail content remains off-limits for Gemini AI training. The recent confusion highlights how sensitive users are to privacy changes — especially when it involves personal emails. Meanwhile, enhanced Gmail notifications and advances in Google’s image generation tech show the company’s broader push to evolve its AI-powered ecosystem.
Read more: Google Unveils Nano Banana Pro: A Powerful New AI Image Generator Supercharged by Gemini 3
