Google has strongly refuted viral claims that it is scanning users’ Gmail messages and attachments to train its Gemini AI model. In response to widespread social media posts and reports, the company stated that these allegations are “misleading”.
The uproar began after a widely shared article from ‘Malware bytes’ fuelled concerns by claiming users could only opt out of this setting by disabling Gmail’s ‘Smart Features’, including spell check.
No change in Gmail settings: Google
A Google spokesperson, Jenny Thomson, told The Verge that the company has not changed any user settings. She emphasised that Gmail’s Smart Features’, such as spell check, predictive text, package tracking, and flight auto-additions to calendar, have been part of Gmail for many years, and they are not used for training Gemini.
Confusion linked to January settings update
Much of the backlash appears to stem from a January update by Google, which separated personalisation settings for Workspace products (Gmail, Calendar, Docs) from other Google applications like Maps and Wallet.
According to Google and independent reporting, some users who had previously disabled Smart Features found them re-enabled, but Google maintains this was not a policy change.
What do Smart Features actually do?
According to Google, when enabled, Gmail’s Smart Features use the content of your emails to personalise your experience, not to feed AI training data.
These features help in tasks such as suggesting replies, tracking orders, and auto-detecting flight bookings.
Turning them off limits personalisation but stops Gmail from processing content for these specific enhancements. Google stresses that enabling them does not give the company permission to feed email text or attachments into model training pipelines, Moneycontrol reported.
Lawsuit against Google sparks concerns
Despite Google’s assurances, a proposed class-action lawsuit filed this month accuses Google of violating California’s Invasion of Privacy Act by secretly enabling its Gemini AI to access private communications across Gmail, Chat, and Meet without explicit user consent.
Google has not publicly commented on the litigation so far.
