Google Expands Data Collection for AI Training: Users Can Opt Out
Google is updating its data retention policies to include a broader range of user-uploaded media for training its artificial intelligence models. The changes, which are rolling out globally over the next few months, mean that images, audio recordings, and videos from interactions with Google Search services may now be saved and used to improve AI performance.
According to a Google support page, saved media will include photos uploaded for reverse image searches (Google Lens), audio from Search Live or Translate speaking practice, voice searches, and other uploaded content. Previously, Google primarily stored text-based search queries.
Users will see a pop-up in their Google Account settings notifying them of the new defaults. The settings, called Search Services History, will be enabled by default for most users. Those who have previously disabled Web & App Activity or Search Personalization may not see the new tab.
To opt out, users can visit the My Activity page, select the Search Services History tab, and uncheck the box next to 'Save media'. Google states that saved media used for training will be 'disconnected' from user accounts and retained for up to four years, even if the original activity is deleted.
Google's move is part of a broader industry trend where companies increasingly rely on diverse data, including audio and video, to enhance AI models. Google's vast user base across multiple services gives it a data advantage over competitors like OpenAI and Anthropic. However, privacy experts note that default opt-in policies place the burden on users to protect their data, which may contribute to privacy fatigue.
In a statement to Wired, Google spokesperson Davis Thompson said, 'These new settings help users get more relevant results and revisit their searches—including visual and voice searches—and they can be turned on or off at any time.'
It remains unclear whether the new settings are enabled by default for all users. Google has not provided a detailed breakdown of the rollout.