Google is trying to calm growing panic about how Gmail data feeds its AI models.
After viral posts claimed the company had quietly changed policies so it could train AI on your emails unless you opted out of “smart features,” Google is now firmly saying that is not what is happening. According to the company, the content of your Gmail messages is not being used to train its Gemini AI model.
In a statement given to The Verge, Google spokesperson Jenny Thomson says recent reports about Gmail settings are misleading and that no one’s options have been changed in the background.
Smart features in Gmail, such as spell checking and automatic suggestions, have been around for years and rely on processing your email data to work, but Google insists that this is separate from training Gemini. In other words, the system can read and act on your messages to offer features without feeding that content back into the big AI training pipeline.
If you do keep smart features switched on, Gmail can do more small helpful things in the background. The setting description mentions that you agree to let Google Workspace use your content and activity to personalise your experience, which covers abilities such as tracking orders, pulling flight details into your calendar, and improving spell check.
Google is now stressing that this wording does not mean your email text is being shipped off to train Gemini or other large models, even if it is analysed locally to power those features.
For now, the situation boils down to a mix of clumsy wording, old features, and new AI anxiety. Google is adamant that Gmail content is not part of Gemini’s training diet. But if you are uneasy about how much your inbox is used to personalise services, it is still worth opening your Gmail and Workspace settings to confirm that the smart feature toggles match what you actually want.
