Google has introduced a new feature in its Search history settings that automatically saves user-uploaded media such as images and voice recordings to improve its AI models. While this setting is enabled by default for most users, Google provides an opt-out option to stop the collection and delete stored activity.
- Google saves images, audio, and video from Search interactions by default for AI training.
- Users must manually opt out and delete saved media to stop data collection.
- Collected data is retained up to four years, even if original activity is deleted.
What happened
Google has updated its Search history options by introducing a setting called Search Services History that stores media uploaded during interactions, including photos used in reverse image searches, voice recordings from Translate, and other audio or video from Search services. This setting is generally enabled by default unless users previously disabled other activity tracking options.
The feature automatically collects this multimedia data to feed Google’s AI training models, with saved media retained for up to four years. Users can opt out by visiting the Search Services History tab on Google’s My Activity page where they can turn off media saving and delete their stored activity.
Why it matters
This update signals a shift towards collecting more diverse and complex data types beyond text queries to enhance AI development. Google’s extensive user base and multiple integrated services allow it to gather a variety of inputs—photos, audio, video—potentially accelerating innovation compared to competitors.
However, the default opt-in approach raises privacy concerns, as users may be unaware their media uploads are retained for years and used in AI training even after deletion of the original search history. Privacy advocates emphasize the need for explicit opt-in consent rather than opt-out models to respect user agency.
What to watch next
Monitor whether Google’s messaging around this default setting evolves to provide clearer justifications and options for users, including more transparent explanations of benefits and risks tied to media data collection for AI training.
It will be important to see if regulators or privacy watchdogs intervene or issue guidelines affecting how companies collect and retain multimedia data for AI, especially in terms of user consent and data retention duration. User response may also influence adoption of alternative search platforms promising stronger privacy protections.