MOUNTAIN VIEW, California, November 22, 2025 – Google has dismissed as “misleading” online reports alleging it scans users’ Gmail emails and attachments to train its Gemini AI models, insisting no policy shifts have occurred and directing users to review existing privacy toggles for smart features like spell-checking.
The uproar began with social media buzz and articles, including a Malwarebytes post suggesting a recent update greenlit email data for generative AI development. A viral X thread amplified the concern, prompting widespread calls to disable features.
Google spokesperson Jenny Thomson responded: “These reports are misleading – we have not changed anyone’s settings, Gmail Smart Features have existed for many years, and we do not use your Gmail content for training our Gemini AI model.”
Gmail’s smart features, introduced years ago, handle tasks such as smart replies, order tracking, and calendar integrations.
For Google Workspace accounts, enabling them permits content use for personalization across the suite, but Google maintains that this stays separate from AI model training.
The company updated these controls in January 2025, allowing users to manage settings independently for Workspace, Maps, and Wallet, as detailed in a product announcement .
Users can access opt-out options via their account dashboard, where toggling off personalization halts data use for those features. A Verge reporter, however, encountered an automatic re-enrollment despite a prior opt-out, highlighting potential glitches.
Google’s support page outlines regional variations: In the EU, UK, and Switzerland, processing draws on contractual needs or legitimate interests, with data retained only as required for service.
The episode reflects persistent privacy jitters in the AI landscape, where firms face scrutiny over data practices amid rapid model advancements.
Google has navigated similar issues, including past EU penalties for ad targeting, but reaffirms that Gmail data remains isolated from broader AI efforts.
Clearer messaging could curb future confusion, but as AI tools embed deeper into daily apps, users must stay vigilant on settings to match their privacy thresholds.
