The photo you snapped of a restaurant menu last Tuesday. The voice search you did in the car. The screenshot you ran through Google Translate because the label was in French. Until very recently, you probably assumed those were just fleeting interactions, gone the moment the result appeared on your screen. They weren’t.
A recent change to Google’s privacy settings allows the company to store your data, including images, files, and audio and video recordings, to improve its AI models. Google is now training its AI models on your images, voice recordings, and videos by default, and a new setting called Search Services History governs this. For most users, it was turned on automatically. Most people had no idea.
The change came via an update to Google’s Search services privacy settings, announced in June via a customer email. Google essentially opted people into this expanded AI training under the guise of giving users more control over their saved history and personalized recommendations. A notification dressed up as a feature improvement is easy to click past. Millions of people did exactly that.
What Changed and What Google Is Actually Collecting
Search Services History is the product of a broader restructuring of how Google manages privacy controls. According to a 2026 report from Computerworld, Google introduced a new Search Services History setting that controls what gets saved and whether it can be used for AI training, alongside a separate Personalized Recommendations setting that controls whether Google tailors results to your account data.
The name sounds fairly innocuous. What it actually covers is considerably broader. According to Google’s support documentation, Search Services History includes your queries, information from sites you visit through Search services, generative AI responses, and your general location. That already describes most of what an active Google user does in a day. But the more significant shift involves media.
This update applies beyond Google Search itself, and also includes other search services such as Maps, Shopping, Flights, Hotels, Translate, and News. When you use Google Lens to search for something visually by snapping a photo, that image may now be saved for AI training. Similarly, if you use voice input in the Google app to search, those audio recordings could be saved.
PYMNTS.com reports that the data collection also extends to any files processed through Google Translate, encompassing images, files and audio and video recordings. So the receipt you photographed to translate the total, the voice note you dictated into a search bar, the screenshot of that contract clause you wanted explained: all of it. Google Lens on a whiteboard, a voice query about a client matter, a contract run through Translate: all of it falls under this setting and can feed Google’s AI development by default.
The current policy is restricted to search-related products; personal repositories such as Google Photos are currently excluded from this specific training data sweep. That’s a meaningful boundary to note, but it doesn’t shrink the scope of what is covered as much as it might sound.
Why This Matters More Than a Typical Settings Update
Calli Schroeder, senior counsel at the Electronic Privacy Information Center, recommends opting out. “Think really hard if you’re comfortable with Google having a picture of your kid, or having a picture of you in a swimsuit that you’re trying to find the maker of,” she told HuffPost. Schroeder also flagged that voice recordings and images of faces raise significant privacy concerns, and that search history has at times been subpoenaed by law enforcement.
People treat their search queries as private thoughts typed into a box. They’re not. They’re records, timestamped and stored, and they’ve been used in legal proceedings before. Adding voice recordings and images of faces to that record expands what law enforcement could one day request and what Google would be obligated to hand over.
Users may want to do a Google AI opt out because their posts, prompts, and files can be used in ways they might not expect. Even though AI systems do not retain information like databases, they can sometimes memorize data and reveal it through sophisticated cyberattacks. Artists and other creators may also want to shield their work from AI training, as the model may learn to mimic their style and reduce market demand for their creations.
Consumer AI products use chats, posts, and account data to train models by default. Once your information is included in a training run, removing its influence from the model is technically complex and not guaranteed: machine unlearning techniques exist, but they are still being refined and are not universally applied by AI providers. The earlier you opt out, the more of your data stays out. Whatever Google trains on today becomes part of the model. You can delete your activity from your account history, but the extent to which that erases a trained model’s learned associations depends on whether and how the provider implements unlearning.
For professionals, the exposure is particularly sharp. If you use a personal Google account for any work-related tasks, you have no protection unless you act. That covers a significant number of people who use their personal Gmail or Google accounts for work side projects, freelance correspondence, or simply because the corporate VPN was being slow that afternoon.
Google and other tech companies are increasingly collecting data that people upload or create when using their services, rather than relying solely on information scraped from the web.
How to Do the Google AI Opt Out Right Now
The process takes about twenty seconds once you know where to look. The catch is that because the rollout is gradual, not every Google account has the new settings interface yet. What you see depends on where your account is in the transition.
If you have already received the new interface, head to the Google Activity Controls page and look for a section called “Search Services History.” From there, use the option to disable the “Save Media” setting within that section, which will stop any media files from being saved and used without eliminating the entire history of things you’ve searched.
If you’d rather eliminate all of your Google Search history from being saved and used even for your own future discovery and recommendations, you can also opt to turn that entire section off. That’s the more thorough option. It means you won’t be able to revisit previous searches or get suggestions based on past activity, but for many people that’s a trade they’re willing to make.
If you don’t see the new section yet, you have two options: you can completely disable all of “Web & App Activity.” Google says if you do this, once your account transitions over to the new approach, all of those “Search Services History” settings will stay off as well. Just be aware that doing so will prevent any and all search history from being saved for you from here on out, which means you won’t be able to revisit your search history yourself and won’t see suggestions and personalization based on past searches throughout Google apps in the future.
If you want to avoid entirely eliminating all of your search history, you can for now uncheck the boxes only for “Include voice and audio activity” and “Include Visual Search History.” That’ll stop search-related media from being saved to your Google account for the time being, though it’s worth setting yourself a reminder to check that same page once a week or so until you see “Search Services History” appear and can confirm that “Save Media” is unchecked as a result of that previous preference.
One important note: turning off Search Services History does not disable Personalized Recommendations, which controls whether Google tailors results to your account data. To turn that off separately, go to Data & Privacy, click Personalization settings, and toggle off Personalized Recommendations in Search services, or go directly to google.com/search-personalization.
What About Gemini?
Search Services History is separate from anything that happens when you use Gemini directly. According to Google’s own documentation, Google collects your prompts, the files you share, and information from apps connected to Gemini. Business customers’ data is not used to train models. Users can keep Gemini from retaining their data, but Google says it will still save chats for 72 hours to respond to you and help keep Gemini safe.
To keep your Google Gemini data out of AI training: go to “Settings & help” on the bottom of the left sidebar when inside the Gemini interface, then look for the Gemini Apps Activity toggle and turn it off. Search Services History does not cover Chrome, Gemini Apps, Google Assistant, or YouTube, all of which save history under their own separate settings. That means a thorough Google AI opt out requires checking several menus, not just one.
The Defaults Are Not Set in Your Favor
The most revealing thing about how this change was rolled out isn’t the data collection itself. It’s the structure of the announcement. Google confirms the media-training use directly, stating in that email to customers: “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.” But a disclosure buried in a permission email, sent to billions of accounts across a gradual rollout, is designed to be consented to without being read.
Google isn’t alone in this. Meta has done the same thing at scale, training its AI on users’ images and media, as well as on content recorded by its AI glasses. The industry logic is consistent: announce a new feature that requires more data, frame the data collection as a personalization benefit, enable it by default, and wait. Most users do nothing. The ones who act have to know exactly where to look.
None of this is a reason to abandon Google entirely. It’s a much more practical argument for treating your account settings the way you’d treat the permissions on a new app. The defaults are not set in your favor. They’re set to collect. The twenty seconds it takes to change that is genuinely worth it, not because it stops everything, but because it stops the most avoidable parts. A photo you took for a quick Lens search is still in there if you haven’t acted. A voice recording from a hands-free search on a Tuesday afternoon is in there. The data that goes into a training run is difficult to remove after the fact. But the data that never goes in can’t be used, either.
AI Disclaimer: This article was created with the assistance of AI tools and reviewed by a human editor.