Google has started rolling out a Search Services History control that lets the company save and use your images, audio, video, and other media from Google Search products to train its generative AI models. The change is separate from your existing Web & App Activity settings, and it’s on by default for many users. If you’ve ever used Google Lens to identify a plant, spoken a voice search, or uploaded a file to Translate, that media could now be retained for up to four years and used to improve AI—even if your account shows it was “disconnected” from the final training dataset.

This article breaks down exactly what’s happening, who’s affected, and the concrete steps you should take right now to protect your privacy, whether you’re a home user, an IT administrator, or a content creator who relies on Google services.

What Actually Changed

On July 12, 2026, Google updated its terms and began rolling out a new “Search Services History” setting across Search, Maps, Shopping, Flights, Hotels, Translate, and News. Within this setting, a sub-toggle called “Save Media” controls whether Google retains your media from interactions. According to Google’s help documentation, “media” includes Google Lens images, files you upload to Search, voice-search recordings, Search Live recordings and transcripts, and even speaking-practice audio from Google Translate.

When the setting is on, Google can use that saved history and media to “develop and improve” its services, including training generative AI models. The company says it disconnects the material from your Google Account before using it for training, filters out sensitive information, and retains it for no more than four years. But the phrase “disconnected from your account” does not mean deleted. Google also warns that turning off the setting only stops future activity from being used; media already saved must be manually deleted, and data already pulled into AI training pools may persist even after deletion.

Crucially, these new controls don’t cover everything. Google explicitly says Gemini Apps, Google Voice, NotebookLM, YouTube, Chrome, Google Assistant, and public posts like reviews fall under other privacy settings or policies. It’s not a one-stop privacy switch.

The setting is still rolling out gradually. If you don’t see “Search Services History” in your Google Activity controls yet, you may still be covered by the older Web & App Activity settings—but you will see the new option soon.

What It Means for You

Home Users

If you use Google’s ecosystem—Android phones, Chrome, Google Photos—you likely have years of search media stored without realizing it. Every time you pointed Google Lens at a receipt or spoke a voice query, that data could now feed into Google’s AI models. Turning off the Save Media toggle is the fastest way to opt out of future contributions. But the real risk is old data: Google may already hold years of your uploads and recordings, and you must manually delete your Search Services History to clear them.

Also watch out for AI Mode. If you’ve connected Google Photos to Search’s “Personal Intelligence” features, Google may use excerpts, summaries, and generated media from your Photos library to train AI. That means vacation snapshots or family videos could influence machine learning models—even if you never shared them publicly.

IT Administrators

The new setting introduces an urgent employee-education challenge. Many workers use personal Google accounts on corporate devices for quick tasks: taking a photo of a whiteboard with Lens, uploading a screenshot to Search, or speaking a translation. That corporate data could end up in AI training pipelines.

Google says education-account Search Services History isn’t used for generative AI training. But for standard consumer accounts—or G Suite accounts where admins haven’t restricted the feature—there’s no blanket safeguard. IT teams should:

  • Audit whether employees access Google Search services on company hardware.
  • Create clear policies prohibiting the upload of work-related media to personal or unmanaged Google accounts.
  • Consider blocking Google Lens, voice search, and file upload features through endpoint management if the risk is high.
  • Check the admin console for any available controls over Search Services History for managed accounts (currently limited; the setting is account-level, not organizational).

Content Creators and Developers

If you use Google’s services to store, translate, or search against your original content, be aware that uploaded files—even those you consider creative work—may be used to train AI. Google says media selected for training is “disconnected from your account” and filtered for sensitive info, but the data can be retained for up to four years. There’s no published mechanism to exclude specific files from that pool once uploaded.

Additionally, the new settings do not apply to YouTube, Chrome, or public posts (reviews, Maps contributions). Those platforms follow their own terms, which may already permit AI training. If you publish content through Google services, read the fine print for each product.

How We Got Here

Google’s hunger for training data isn’t new. The company has long collected user activity to refine products, but the generative AI boom accelerated the need for multimodal data—images, audio, video. In 2023, Google updated its privacy policy to broadly state it could use publicly available information to train AI. The shift to explicitly saving and using personal search media is a logical next step, but it arrives amid heightened regulatory scrutiny.

In July 2026, the European Union opened an investigation into whether Google imposes unfair terms on content creators for AI, as reported by PC Gamer. This new Search Services History setting seems designed, in part, to give users a clearer opt-out while still preserving Google’s access to a massive training corpus. By creating a separate history category, Google can argue it’s offering more granular control than burying the same data under the broad “Web & App Activity” umbrella.

But privacy advocates note that the default setting is on, and users must discover and navigate nested menus to disable it. The rollout also coincides with Google’s push of AI Mode and Personal Intelligence features, which encourage users to connect more data sources into Search—further blurring the line between personal storage and AI training material.

What to Do Now: A Step-by-Step Guide

Check and adjust your settings immediately. Here’s how:

  1. Open your Google Activity controls: Go to myactivity.google.com and sign in.
  2. Locate “Search Services History.” If you see it, click on it. If not, look for “Web & App Activity”—the new settings may still be rolling out.
  3. Inside Search Services History, toggle off “Save Media.” This stops Google from saving your images, audio, video, and other media from future interactions. You can also turn off the entire Search Services History, but note that may disable other features you rely on (like personalized search).
  4. Delete existing history: Still on the Search Services History page, find the option to delete activity (usually a trash icon or “Delete” button). Choose “Delete all time” to wipe past media. Google says this prevents that data from being used for AI training going forward, though any data already selected for training may remain retained.
  5. Review connected apps: In the Google app, tap your profile icon > “AI Mode” or “Personal Intelligence.” If you see Google Photos or any Workspace apps connected, and you don’t want their contents used for AI, disconnect them. Remember: Google may still use already-processed summaries or excerpts even after disconnection.
  6. Check other services: This new setting does not cover Gemini, Voice, NotebookLM, YouTube, Chrome, or public contributions. Visit each service’s privacy controls separately and dial back data sharing where possible.
  7. For IT administrators: Distribute clear guidance to employees. Share screenshots of where to find and disable the Save Media toggle. Remind staff not to use personal Google accounts for work. Monitor for organizational policy updates—Google may eventually surface admin-level controls for Workspace accounts, but currently, it’s an individual-user setting.

Outlook: What to Watch For

Google’s move is part of a broader industry trend: companies are building more explicit, but often opt-out, controls for AI training data. As EU regulators dig into Google’s terms for creators, we might see forced changes, such as an opt-in default. Microsoft, too, has faced pushback over how it uses Bing searches and Windows data to train Copilot.

In the near term, expect Google to refine these controls as user backlash and media coverage grow. A likely next step: an “opt-out from AI training” button that’s less buried. But until then, the burden remains on you to lock down your settings. The safest posture? Assume anything you upload, speak, or snap through a Google service can become training fodder—and act accordingly.