Google recently adjusted the privacy settings of the search service to allow users to upload or generate images, audio, video and files in some products, which may be saved and used to improve their AI models. The changes were notified to users by mail in June, but the new settings are default-opened and coverage is not limited to Google Search.
Cover multiple search-related services
This adaptation applies to a number of products within the Google Search Service, including Maps, Shoping, Flights, Hotels, Translate and News. In the case of Google Lens, the user takes a photo and initiates a visual search, which can be saved.
Sound recordings may also be retained if users use Seach Live for voice searches in Google applications. When speaking practice is performed using Google Translate, the relevant voice may also be saved.
Google has indicated in its e-mails to users that the stored media content will be used, like search history, to develop and improve Google services and technologies, including AI models and security measures. Its help documentation also indicates that historical records can be used to train the generated AI model and be used for safety protection with manual clearance.
New settings split with old options
Prior to this update, users managed the preservation of historical search data mainly through Web and Applications. After updating, Google splits the original settings into two parts: The other part is the new search data setup, which is started by default.
This means that even if the user has previously adjusted the data retention of Web with the application activity, this does not mean that the new data saving options in the search service are simultaneously closed. The search for the relevant media content has been transformed into a stand-alone project.
- Save Search Service History
- Whether to save media content separately
- Time period for automatically deleting data
Google provides automatic elimination cycles of 3 months, 18 months and 36 months.
Industry is expanding user data collection
These adjustments also reflect common trends in AI competition among large technology firms. In addition to capturing public web-based information, the Platform is increasingly using content uploaded, generated or recorded by users in their home services as a source of model training data.
TechCrunch reported that Meta was also using user pictures, media content and an AI glasses-recorded data training model on a large scale. For Google, this change was not only related to individualized recommendations and advertising, but also expanded the AI training-accessible data pool.
For users, the effect is mainly on changes in default settings. In the future, parts of search, voice and image-related services may continue to be saved and used for AI training if they are not voluntarily closed.