In short
Google is rolling out new Search privacy settings that can store images, voice, files and other media for AI training unless users opt out. The change places the responsibility on users to find the setting in My Activity and disable it if they want to limit data use.
- Google is expanding Search data retention to include images, audio, files and voice interactions.
- The Search Services History feature may be enabled by default for some users.
- Users can opt out through Google My Activity and should also uncheck the separate media-saving control.
- Privacy experts say the change adds to the growing burden of constant opt-outs across tech platforms.
- The update highlights how AI development increasingly depends on large-scale, multimodal user data.
Google is rolling out a significant change to Search that gives the company broader access to the media people upload or generate through its search tools — and it is doing so in a way that places the burden on users to notice, understand and disable the feature if they do not want their data used for AI training.
The update, communicated to users through a privacy settings email and account notices, changes how Google handles saved Search Services History. The setting covers more than typed queries. It can include images, files, voice searches and audio or video recordings created through Google Search products such as Lens, Translate speaking practice and Search Live. In effect, Google is expanding the kinds of interactions it can retain and later use to improve its AI systems.
For many users, the most striking part of the rollout is that the setting may already be turned on when it appears in their accounts. That means people who do not actively check their privacy dashboard could be opting into AI-related data retention without realizing it.
What Google changed
At the center of the rollout is a new Search Services History control in Google’s account settings. The feature is part of a global launch that the company says will arrive over the coming months. Once available, it gives users a place to review what Google stores from Search activity, turn the setting off entirely and delete past records.
But the control is not limited to traditional search history. Google now describes the stored material more broadly, including media people upload, recordings tied to voice interactions and content produced through tools such as Google Lens and Translate practice sessions. That is an important shift because it means the data pipeline feeding Google’s systems is no longer just text-based query logs.
In practical terms, the change reflects how modern AI development works. Large models are improved not only with words, but also with multimodal inputs like images, audio and video. The more varied the data, the more opportunities companies have to train products that better understand how people search, speak and interact across formats.
What falls under Search Services History
Google’s description of the setting suggests the scope is wider than many users may expect. Rather than only preserving search terms, the company says saved media can include the following:
- Images uploaded to Google Search tools, including Lens
- Files shared through Search-related services
- Voice searches
- Audio recordings from speaking practice or live search features
- Video and other media tied to certain search interactions
This matters because a user might think of Search as a simple query tool, while Google is increasingly treating it as a rich data source that captures gestures, speech and visual context.
Why the default matters
The question that will likely concern privacy-minded users most is not just what the setting does, but how it appears to be configured. The feature was active by default in at least one test account reviewed for the rollout, unless earlier account-wide privacy controls had already been disabled. That raises an obvious issue: many people may not know they are allowing Google to retain more of their search-related media for longer-term use.
Google did not directly answer a question about whether the setting is enabled by default for all eligible users. In its written response, the company said the new controls are intended to offer people more flexibility over saved history and to make it easier to revisit past searches, including visual and voice searches. Google also said users can turn the feature on or off at any time.
Still, the default state matters because privacy choices depend on attention, time and understanding — three things many users lack when settings are introduced through account notices or buried in menus. In that context, an opt-out model can function less like a choice and more like a quiet permission slip.
Google said the new controls are meant to give users “more control over saved history” and to support features like revisiting earlier visual searches or continuing a Search Live conversation. The company did not explain why the feature appears to be switched on for some accounts by default.
How to turn it off
Users who want to avoid having their search-related media used for AI training can manage the setting through Google’s My Activity page. The process is not especially difficult, but it does require users to know where to look.
Once the feature is available in an account, users can open My Activity, select the Search Services History section and review the controls there. From that page, they can disable the overall setting and delete stored activity. There is also a separate checkbox related to saving media, which users need to uncheck if they want to prevent image uploads and other media from being retained for AI training.
That distinction is important. Disabling one part of the feature does not necessarily mean every data-retention element is gone. Users need to make sure the media-saving box is also switched off if they want to limit how their uploads are handled.
| Action | Where to find it | What it affects | Why it matters |
|---|---|---|---|
| Review Search Services History | Google My Activity | Shows what Search-related data Google stores | Lets users see the scope of retention |
| Turn off the feature | Search Services History tab | Stops the broader history setting | Limits ongoing collection and retention |
| Uncheck “Save media” | Within the same settings page | Prevents saved uploads from being used for training | Crucial for users who do not want image or audio data included |
| Delete existing activity | My Activity controls | Removes past saved records | Reduces the amount of data already stored in the account |
What Google says it gains
Google’s public explanation for the change is straightforward: storing more of this material can make search experiences more useful. The company says saved history helps people revisit previous searches and continue previous interactions. That includes features such as visual lookups and conversational search experiences.
From a product perspective, this makes sense. Search is no longer limited to text strings entered into a box. Users now search by snapping photos, uploading screenshots, speaking aloud and practicing pronunciation. If Google can connect those inputs to a user’s account history, it can build more personalized and potentially more accurate systems.
From an AI perspective, the incentive is even clearer. Data is fuel. The company that has access to more of it, and more varied forms of it, has a better chance of refining its models and staying competitive. In a field where model performance can hinge on scale and diversity, Google’s ecosystem gives it a major advantage.
Multimodal AI needs multimodal data
One reason the policy change matters is that AI training increasingly depends on different kinds of data at once. Text remains important, but language models and related systems are now often paired with vision, audio and video capabilities. That means companies want to learn from a broader range of user interactions.
Google is uniquely positioned here because its products already touch many parts of users’ digital lives. Search, Lens, Translate and voice-based tools all produce different kinds of input that can be connected to the same account ecosystem. That breadth is valuable for training, but it also raises the stakes for privacy.
A privacy advocate at the Electronic Frontier Foundation said Google stands apart from many competitors because people have used its services for so long that they may be less likely to question the amount of data being collected.
Why privacy experts are uneasy
The rollout has prompted concern not because Google is doing something unprecedented, but because it reflects a broader pattern that has become standard across the tech industry: users are often expected to opt out of data use rather than actively agree to it first.
That approach can be especially frustrating when the data in question is not obviously sensitive at first glance. A voice query, a photo search or a translation practice session may feel routine in the moment, but together they can reveal a lot about a person’s habits, interests, location and behavior.
Privacy specialists say that forcing users to navigate yet another settings page places too much responsibility on them and too little on the company collecting the information. It also compounds the fatigue many people already feel from repeated prompts, consent banners and default-enabled data features across different apps and platforms.
The cost of constant opt-outs
There is a psychological dimension to this debate that goes beyond policy. When users are repeatedly asked to find and disable new data-collection features, some eventually stop trying. The sheer volume of choices can create a sense of helplessness.
That matters because privacy is not only about legal rights or technical controls. It is also about whether ordinary people can realistically understand and exercise those rights. A setting that is technically available but practically hidden may satisfy a formal checklist without giving users meaningful control.
An AI and privacy advocate at the Consumer Federation of America said these changes force people to do additional mental work each time a familiar service changes its rules, effectively making them decide whether they still feel comfortable using tools they have relied on for years.
That burden may seem minor in isolation, but across dozens of services it adds up. Over time, many people stop evaluating each new policy update carefully and simply continue using the product. That inertia is part of what makes large platforms so powerful.
What this says about Google’s AI strategy
The new settings are about more than privacy. They also offer a glimpse into how Google is preparing its core products for the AI era. Search is one of the company’s most important surfaces, and the data it collects there can help inform not just rankings and relevance, but also model training, personalization and new product development.
Google’s challenge is to grow its AI capabilities without alienating users who are increasingly wary of how their information is being used. The company is trying to present the change as a convenience feature and a control feature at the same time. It is telling users they can revisit past content more easily while also quietly expanding the data pool that supports AI development.
That dual message is common in the current tech landscape. Companies often describe new data practices in terms of usefulness, memory or customization, while the long-term training value is mentioned in much smaller print. Users then have to decide whether the trade-off is worth it.
Convenience versus consent
The central tension here is familiar: personalization can be helpful, but the more a service knows, the more it can shape and potentially expose user behavior. For many people, the convenience of easily recovering old searches or continuing a conversation with an assistant may outweigh the privacy downside. For others, the risk of data reuse is not acceptable.
What makes the current rollout notable is that it blends both propositions into one feature. The user-facing pitch emphasizes recall and control. The backend implication is data retention for AI training. Those two ideas are not necessarily incompatible, but they do require transparency if users are expected to make informed decisions.
How users should approach the change
If you use Google Search features that involve images, audio or voice, this is a setting worth reviewing as soon as it appears in your account. Waiting may mean more of your data is retained before you notice the feature is active.
At minimum, users should check the Search Services History page in My Activity, review what kinds of content are being stored and decide whether they want media retention enabled. If the answer is no, they should disable the setting and clear existing activity as soon as possible.
It is also wise to revisit related Google privacy controls. Search history is only one part of the broader data ecosystem. Account-level choices about Web & App Activity, Search Personalization and location history can all affect what Google stores and how it uses that data across services.
Practical steps to review
- Open Google My Activity in your account settings.
- Look for Search Services History.
- Check whether the setting is enabled.
- Review the “Save media” option separately.
- Turn off any controls you do not want active.
- Delete existing stored activity if you want to reduce retained records.
The broader industry pattern
Google is not alone in expanding the ways it can use consumer data for AI. Across the tech sector, companies are looking for more training material, better personalization signals and richer context from real-world behavior. That makes privacy controls a frontline issue, not an afterthought.
As AI products become more capable, the pressure to feed them more data will only intensify. That creates a recurring tension: the tools people enjoy most are often the ones most eager to learn from them. The result is a marketplace where convenience and surveillance can become difficult to separate.
For Google, the new Search settings are a reminder that the company’s data advantage remains central to its AI ambitions. For users, they are a reminder that privacy protections often depend less on dramatic policy battles than on careful attention to account dashboards and buried toggles.
That may not be a satisfying answer. But in the current AI era, the difference between keeping and sharing your data can come down to a single checkbox.
What users should remember
The key takeaway is simple: Google’s Search products are now collecting and retaining a wider range of media than many users may expect, and some accounts may see the feature enabled automatically. If you do not want images, audio, files or voice interactions tied to this new training pipeline, you need to actively check your settings.
For people who value convenience and personalization, the feature may seem like a fair trade. For others, it is another example of how AI progress is increasingly built on user data that is easier to collect than to reclaim.
