On July 6, 2026, TechCrunch reported on what it described as an “under-the-radar update” from Google. A June 2026 policy change quietly made the images you send through Google Search — photos run through Lens, audio from voice queries, files uploaded to Google Translate — fair game for training AI models. Most users had no idea it happened.
The change introduced a new setting called Search Services History. It was turned on by default. And critically, it is separate from Web & App Activity, the setting most privacy-conscious users had already turned off. Opting out of one did not opt out of the other.
This matters because Lens usage has exploded. Hundreds of millions of people use it to identify plants, translate menus, search for products, and look up items using their camera. Every one of those photos was now, by default, contributing to Google’s AI training pipeline.
What “Search Services History” Actually Captures
The new setting covers media you send to Google through Search-related services. According to reporting from AndroidHeadlines and PhoneArena, that includes:
- Photos run through Google Lens for visual search
- Voice recordings from voice-activated queries
- Files uploaded to Google Translate
- Other media inputs sent to Search-adjacent tools
What it does not include — and Google has been explicit about this — is content you have stored in Google Photos. Existing photos in your personal library are governed by a separate policy and are excluded from this particular change.
The issue is that Lens is often the bridge. You take a photo of something and Lens processes it through Search to give you results. That workflow, which feels like using your camera, is technically sending media through Search Services, and it falls under the new setting.
Why Your Existing Opt-Out No Longer Works
Many people reading this have already taken steps to limit Google’s access to their data. Turning off Web & App Activity is the advice that privacy guides have given for years. If you followed it, you may assume you are already protected.
You are not — or at least, not for this.
Search Services History is a new, distinct toggle. Google added it to Google Account settings in June 2026, and it was enabled automatically. The previous Web & App Activity setting controls your search history in the traditional sense. The new Search Services History setting specifically targets media: the images, audio, and files you transmit to Google’s search tools.
To confirm whether this applies to you: navigate to myaccount.google.com, select Data & Privacy, find the section labeled “History settings,” and look for Search Services History. If it shows as active, your media is currently being collected for AI training.
How to Turn It Off
The opt-out is straightforward once you find it. Here is the full path:
- Go to your Google Account at myaccount.google.com
- Select Data & Privacy in the left sidebar
- Scroll to History settings
- Click on Search Services History
- Toggle it off, or uncheck Save Media specifically
The “Save Media” checkbox is the one that controls AI training use. You can turn off the entire history, or selectively uncheck media while leaving text search history intact — your choice.
After you turn it off, Google will no longer save or use new Lens queries and voice searches for AI training. Existing data that was collected before you opt out may still be retained according to Google’s data retention schedule, which you can manage separately under the same Data & Privacy section.
The Broader Pattern This Represents
This is not the first time Google has used a default-on setting to enroll users in something most of them would not have agreed to if asked directly. It follows a well-established playbook: introduce a new data collection category with a new toggle, make it active by default, and wait for the small percentage of users who read privacy blogs or notices to find it.
The June 2026 email that Google sent to users was, by most accounts, easy to overlook. The setting quietly appeared. TechCrunch flagged it a month later. By that point, a month of Lens queries and voice searches had already been processed.
Google’s explanation is that this training makes search results more accurate and relevant over time. That may be true. But the question of whether you want your photos and voice used for that purpose is yours to answer, not Google’s to assume.
What This Means If You Use Lens Regularly
For casual users — people who occasionally use Lens to identify a plant or scan a QR code — the practical exposure is low. The images are mundane, and the training use is unlikely to cause direct harm.
For people who use Lens as part of a regular workflow — scanning receipts, translating documents, photographing personal correspondence, or researching medical symptoms — the picture is different. Those images may contain sensitive context: financial information, health data, personal relationships, private locations.
Lens is often positioned as a convenience feature. It does not feel like submitting data to a training pipeline. But that is precisely what it became in June 2026, and the default position assumed your consent.
The Larger Question of Photo AI Defaults
The timing of this change is not accidental. Google is in the middle of a significant push to integrate AI across its products, from Gemini to Google Photos’ Memories feature to Search Generative Experience. All of those require training data, and personal photos and media are some of the richest training signal available.
The problem is that “rich training signal” is the same thing as “sensitive personal data.” Photos from your camera contain contextual information about your life — where you have been, what you own, who you are with, what you are researching. When that data flows into AI training without explicit opt-in, the asymmetry between what Google gains and what users give up is significant.
Other platforms have taken similar steps in recent years. Meta updated its terms to use public social media content for AI training. Adobe’s policies drew outrage when users noticed training clauses in updated terms. The pattern repeats: companies with large stores of user-generated content find ways to use it for model training, often with the minimum disclosure legally required.
What a Private Alternative Looks Like
The strongest protection here is not about fine-tuning Google’s settings — it is about which services receive your photos in the first place.
For photos you want to keep private — family images, personal documents, sensitive snapshots — the question is whether the platform receiving those files has made an explicit, unconditional commitment not to use them for AI training.
Google’s commitment on this is complicated by the June 2026 change: they do protect Google Photos content, but Lens-routed images are now fair game unless you opt out. Many users will not make that distinction, and the default setting works against them.
Services that treat “your files are never used for AI training” as a core product promise — not a secondary setting buried in account preferences — offer a fundamentally different model. That commitment needs to be in the product design, not the settings panel.
Action Checklist
If you use Google Lens, Google Search with voice, or Google Translate with uploaded files:
- Log into myaccount.google.com
- Navigate to Data & Privacy → History settings → Search Services History
- Toggle off Search Services History, or specifically uncheck Save Media
- Review your Web & App Activity setting separately — it covers text-based search
- Check Google Photos settings for the AI-based Memories and Smart Albums features and adjust to your comfort level
- Consider which photos and files you route through Google services versus keeping in a private storage app
The opt-out takes two minutes. The default, left unchanged, is indefinite enrollment in AI training.