Google denies viral claim that Gmail uses your emails to train its Gemini AI model

Google has strongly pushed back against viral claims that Gmail is using personal emails and attachments to train its Gemini AI. According to a Google spokesperson, Jenny Thomson, who told The Verge that these reports are misleading," said, "We do not use your Gmail content for training our Gemini AI model.” He added that Gmail smart features have existed for years, like spellcheck, predictive text, package tracking, and calendar integration have existed for years and using them does not mean Google has access to your personal emails.

Google smart feature vs AI training

The confusion has arisen since a report by the platform Malwarebytes, which published a finding claiming that Google was analysing private emails and attachments to improve AI tools like Smart Compose and Smart Reply. In January, Google introduced a toggle option for Workspace smart features and another for products like Maps and Wallets. Since there are two separate toggles for smart features users, they may mistakenly opt in one and not realise. The possibility that there is a risk that settings may be misunderstood or re-enabled is a huge red flag. Google claims smart features do analyse content locally or in a limited way to “personalise your experience across Workspace”, but this is not similar to feeding your correspondence to Gemini. This explanation, however, did not address all issues. On November 11, a class action lawsuit was filed against Google, alleging that it is tracking consumers' communications without consent after it secretly turned on’ Gemini AI for all users.