Google refutes allegations of Gmail data being used for AI training

Google is taking a stand against false information circulating on social media and websites, such as a recent article from Malwarebytes, alleging that Google has altered its policy to utilize Gmail messages and attachments for training artificial intelligence (AI) models. The claim suggests that the only way to opt out of this practice is by disabling “smart features” like spell checking.

However, Google’s spokesperson, Jenny Thomson, clarified to The Verge that these reports are inaccurate. Google asserts that they have not made any changes to users’ settings, emphasizing that Gmail Smart Features have been in place for many years and that they do not utilize Gmail content for training their Gemini AI model.

Despite this clarification, users are advised to review their settings. A Verge staff member shared an experience of opting out of certain Smart Features only to find themselves automatically opted back in. Google recently updated its smart feature personalization settings, allowing users to disable features individually for Google Workspace and other products like Maps and Wallet.

Enabling Gmail’s smart features offers conveniences such as spell checking, order tracking, and effortless integration of flight details from emails to calendars. While opting in to these features within Workspace grants Google permission to personalize the user experience using Workspace content and activity, Google clarifies that this does not involve using email content for AI training purposes.