Google Search's Default AI Training Feature Now Includes Photos, Audio, and Video

Google Search Services History Defaults to Media Retention

According to Wired AI, Google is rolling out a new data-handling policy called Search Services History that automatically retains users’ uploaded media—including images, audio files, and video recordings—for use in training AI models. The feature is enabled by default for most users and will reach all accounts over the coming months. Davis Thompson, a Google spokesperson, confirmed the rollout but did not respond to questions about the opt-out-required design.

What Data Gets Saved and How Long It Stays

The policy captures significantly more than text-based search queries. According to the feature’s description cited by Wired AI, saved media encompasses “images, files, and audio and video recordings from your interactions with Search services”—including Google Lens photographs, voice-search audio, Translate speaking-practice recordings, and user-uploaded content. Google stores this training data for up to 4 years, Wired AI reports, even after users delete the original activity from their accounts. Once media is used to train AI models, Google disconnects it from the user’s account, making retroactive deletion impossible.

How to Opt Out

Wired AI provides a clear pathway for users seeking to prevent their media from entering AI training datasets. Users should navigate to Google’s My Activity page, select the Search Services History tab, and uncheck the “Save media” box. The same page allows users to review what Google has retained from their Search history and disable the entire Search Services History feature. Wired AI emphasizes acting quickly: once data enters the training pipeline, users lose control over its use and retention.

Why This Matters

This rollout exemplifies a broader pattern in which tech platforms shift data-collection defaults toward broader AI training purposes without explicit, upfront user consent. By requiring opt-out rather than opt-in, Google ensures maximum enrollment in the training program—many users never visit privacy settings or remain unaware of the policy. The 4-year retention window for training data, even after account-level deletion, creates a persistent privacy liability that users cannot undo after the fact. Teams evaluating search-platform privacy policies and compliance obligations should audit their Google Search usage and account settings immediately, particularly in jurisdictions with strict data-retention or cross-use restrictions (GDPR, state privacy laws). For individual users, this update underscores the importance of proactive privacy management in ecosystems where defaults are increasingly aligned with corporate AI development priorities rather than user preferences.