On the surface, Google's latest privacy update looks like a gift to users — more granular controls, clearer settings, and a cleaner separation between history management and personalization. But for anyone paying close attention to what's actually changing under the hood, the rollout raises a question that deserves serious scrutiny: why is the new media-saving feature switched on by default?
What Google Is Actually Changing
The company is in the process of splitting its longstanding Web & App Activity framework into more targeted controls. Going forward, users will interact with two distinct settings for Search services:
- Search Services History — governs whether Google saves your activity from Search, Maps, Shopping, Flights, Hotels, Translate, and News to your Google Account.
- Personalized Recommendations — controls whether Google uses that saved data to tailor what you see across its services.
Google Play is receiving parallel treatment, with new Play History and Personalization in Play settings rolling out simultaneously — even for users who have never actively used the platform.
The separation of history-saving from personalization is, in principle, a meaningful improvement. A user who wants to revisit past searches for convenience but doesn't want algorithmic targeting should theoretically be able to have one without the other. That's a legitimate privacy distinction, and it's worth acknowledging.
The Detail That Deserves a Second Look
Buried within the transition is a subsetting called Save Media — and this is where the story gets more complicated.
According to Google's own communications to users, this feature expands what gets captured under Search Services History to include images, audio files, video, and other files generated through interactions with Search services. That means visual queries submitted through Google Lens, voice interactions with Search Live, and other multimedia inputs can now be stored in your account history.
"Like your Search Services History, your saved media is also used to develop and improve Google services and technologies, including AI models and safety measures."
— Google, in communications to users
For users who already had Web & App Activity enabled, this Save Media subsetting will be automatically turned on during the transition. That's an opt-out model, not opt-in — a distinction that matters enormously from a privacy standpoint.
Why This Matters Strategically
Google frames media-saving as a quality-of-life feature — enabling users to revisit Lens searches or pick up a voice-based session where they left off. That use case is genuine. But the secondary application — using that media to train AI models — deserves transparent, prominent disclosure, not a line item in a transition email.
Multimedia interaction data is substantially richer than text-based search history. Images, voice recordings, and video snippets carry contextual signals that text cannot replicate: the environment in which a photo was taken, the acoustic characteristics of a voice, the visual content of a document someone photographed. The aggregation of this data, at scale, represents a meaningful expansion of Google's behavioral dataset.
This comes at a moment when regulators on both sides of the Atlantic are scrutinizing how large technology platforms collect and leverage personal data for AI development purposes. The EU's AI Act and ongoing enforcement actions under GDPR create a complex compliance backdrop for features that default to data collection for model training.
What You Should Do
Whether you're an individual user or a security-conscious professional advising an organization, the immediate action steps are straightforward:
The Broader Pattern
This update follows a recognizable pattern in how major platforms evolve their data practices during periods of rapid AI development. Features that benefit the user are foregrounded; features that benefit the platform's training pipelines are embedded in the defaults. The controls are real — Google is not being deceptive about their existence — but their placement in the flow assumes a level of active attention that most users will not apply.
True privacy empowerment doesn't just provide controls. It ensures those controls are encountered at the moment of consequence, with clear language about what's at stake. A transition email with opt-out defaults doesn't fully meet that standard.
The good news is that the new architecture, if configured deliberately, does offer more precision than what came before. Whether users take advantage of that precision is another matter entirely.