A recent update to Google’s privacy settings, quietly announced in June via customer emails, has expanded the company’s ability to collect and store user data, including media files such as images, audio, and video recordings. This data collection supports the training of Google's AI models, meaning that any media uploaded to Google services like Search, Maps, Shopping, Flights, Hotels, Translate, and News may be used to improve AI technologies by default. Features like Google Lens, voice searches including in the new Search Live mode, and Google Translate contribute media that Google can retain and analyze for AI development unless users explicitly opt out.

This change introduces two new user controls: Search Services History and Personalized Recommendations. These settings allow users to regulate how their search activity and uploaded media are stored and used for AI training and personalization. Unlike earlier configurations that bundled search data under “Web & App Activity,” Google has now made data related to Search services a separate opt-in area, which remains enabled by default. Additionally, Google provides options for users to determine how long their saved data is retained, with choices ranging from three months to three years.

For those concerned about privacy and data usage, Google offers straightforward ways to opt out of having their media and search history used for AI training. Users can adjust settings on the Search Services History page to disable the saving of media separately from general search history and set automatic deletion timelines. Further privacy controls are accessible through the Google My Activity page, where users can manage other types of saved data, such as location history and YouTube activity.

This move by Google reflects a broader industry trend where tech companies increasingly rely on user-generated data to refine AI capabilities. Meta, for example, also uses consumer photos and data captured by its AI glasses for AI training at scale. While this shift aims to enhance AI models, it raises significant privacy questions, especially since many users might be unaware that their content is being utilized unless they actively change their settings.