Google is refuting a wave of social media claims accusing the company of secretly using Gmail messages to train its Gemini AI model. The tech giant says the allegations—widely shared across X and repeated in a Malwarebytes blog post—are simply not true.
In a statement to The Verge, Google spokesperson Jenny Thomson dismissed the rumors as “misleading,” stressing that the company has not altered user settings and does not use Gmail content to train Gemini. She clarified that Gmail’s Smart Features, which help users manage emails and suggestions, have been around for years and operate independently of the Gemini training pipeline.
Despite this, online backlash grew as users speculated that private emails and attachments were being fed into AI systems without consent. One post called the situation “the largest consent manufacturing operation in history,” while others warned users to manually disable Smart Features—incorrectly claiming they enabled AI training by default.
Google’s policies paint a different picture. Under the Google Workspace Privacy Policy, personal Workspace content—such as emails, Docs, and Sheets—is not automatically accessed by Gemini and is not used to train the model. The only time Gemini can process Workspace data is when a user explicitly asks it to, such as requesting the AI to edit a document.
This is the second major rumor targeting Gmail in recent months. Earlier this year, viral posts falsely suggested that Google had issued a universal password-reset warning to all Gmail users, prompting widespread confusion among its 2.5 billion global users.
While the latest claims have been debunked, concerns about tech companies’ use of personal data remain widespread—especially in jurisdictions with stronger privacy regulations. Companies like Meta and LinkedIn have recently announced that portions of user data may be used to improve AI systems, stoking fears of overreach.
Google has also faced scrutiny before. In May 2025, the company agreed to a $1.375 billion settlement after being accused of collecting Texans’ biometric data without proper consent—an incident critics say contributes to ongoing distrust.
As misinformation spreads quickly online, Google insists that Gmail data remains off-limits for Gemini training unless users knowingly share it themselves.