Google is fighting back against viral claims spreading across social media that it changed its privacy policy to use Gmail messages for AI training. The company tells The Verge these reports are "misleading" and emphasizes it doesn't use Gmail content to train Gemini. But some users report being mysteriously re-enrolled in smart features they'd previously disabled, adding fuel to the privacy concerns.
Google finds itself in damage control mode today as viral social media posts and cybersecurity articles claim the company quietly changed its privacy policy to harvest Gmail messages for AI training. The allegations, which spread rapidly across platforms including a viral Twitter thread and a Malwarebytes report, suggest users must disable "smart features" like spell checking to prevent their emails from being used to train AI models.
"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," Google spokesperson Jenny Thomson told The Verge in a direct response to the growing controversy.
But the story isn't quite that simple. At least one Verge staffer discovered they'd been mysteriously re-enrolled in Gmail Smart Features despite previously opting out, adding a wrinkle to Google's denial. This suggests something changed in how these settings are managed, even if the underlying functionality remained the same.
The confusion stems from Google's January update to its smart feature personalization settings, which now allows users to control Google Workspace features separately from other Google products like Maps and Wallet. This administrative change appears to have reset some users' preferences, creating the impression of a policy shift.
Gmail's smart features, which have been around for years, power conveniences like automatic flight additions to your calendar, package tracking, and enhanced spell checking. When enabled in Workspace, users "agree to let Google Workspace use your Workspace content and activity to personalize your experience across Workspace," according to the settings page.
It's that language - "use your Workspace content" - that's triggering privacy alarms. In an era where every major tech company is scrambling to train increasingly sophisticated AI models, users have good reason to be suspicious of any data collection that wasn't explicitly discussed before.
