Google denies analyzing your emails for AI training (www.zdnet.com)

Google says it did not secretly scan users’ private emails to train its Gemini models after a proposed class-action suit alleged the company quietly enabled AI access to Gmail, Chat and Meet. A Google spokesperson denied any change to users’ settings and said Gmail content isn’t used to train Gemini; Malwarebytes later revised its reporting, blaming a rewording and relocation of Gmail’s “smart features” controls for widespread misunderstanding. Security researchers note Gmail does scan message content to power longstanding features (spam filtering, categorization, Smart Compose/Reply), but that’s different from using that data to build AI models. The technical crux: three auto-enabled settings control personalization and “smart” features—“Turn on smart features in Gmail, Chat, and Meet,” “Smart features in Google Workspace,” and “Smart features in other Google products.” Enabling them lets Google surface calendar events, provide personalized search and suggestions, and enables Gemini-assisted Workspace actions like summarization or draft creation. Concern centers on why these are on by default and how visible the controls are. Users can opt out (Gmail > Gear > View all settings > General > Smart features, then manage Workspace smart features) on desktop and mobile, but doing so may disable conveniences like Smart Compose and Smart Reply.