Google has once again revised its data usage policy, and this time it comes with broader implications for user privacy. As of last month, the company now automatically includes any media you upload to its search services—such as images, videos, and voice recordings—in the training data for its large language models (LLMs). Unless you manually opt out, every photo you snap with Circle to Search, every voice command you issue through Search Live, and every video you share becomes fodder for improving Google's generative AI. The change was announced via email and quietly implemented, catching many users off guard.

What does the new policy really mean?

The updated documentation states: "You can choose whether Google saves media to your Search Services History when you're signed in. Saved media includes your images, files, audio, and video from your interactions with Search services. Your media may be used to improve your experience on Google services, like letting you revisit your past visual searches. Saved media may be used to develop and improve Google's AI models and technologies, as well as the Google services that use them."

In plain language, this grants Google permission to train its AI on your personal media without asking for explicit consent beyond the initial email notification. Because the setting is enabled by default, millions of Google users are now contributing to LLM training without even realizing it. The scope is broad: anything you capture with your phone's camera during a search, any voice snippet used to perform a Google Translate pronunciation check, or any video you upload as part of a visual query can be stored and repurposed.

Privacy risks you should know about

While Google frames this as a way to improve search relevance and AI capabilities, there are tangible privacy concerns. For example, consider a scenario where you chat with a Google AI assistant about leaving your job. You might share the name of your company, your boss's identity, and your reasons for wanting to resign. That conversation—including all the personal details—may be used to train the same LLM that your employer queries weeks later. The result? Your boss could encounter a response that includes fragments of your private discussion, effectively doxxing your plans.

Another risk involves voice data. If Google uses your voice recordings to train its models, someone could potentially reconstruct a synthetic audio clip of you saying something you never uttered. This is not science fiction; voice cloning technology is advancing rapidly, and feeding it with real samples makes it even more accurate. Similarly, your images and videos could be used to improve facial recognition or object detection algorithms, raising questions about biometric data exploitation.

Google's move is part of a broader industry trend where tech giants push the boundaries of data usage to fuel AI development. As users generate more multimedia content than ever before—over 1.8 billion images are uploaded daily across platforms—companies see an opportunity to harvest that data for training. The difference here is that Google's policy change applies retroactively, meaning any media you've previously stored in your search history could already be in the training pool.

How to opt out: step-by-step guide

Fortunately, you can regain control over your data by adjusting a few settings. Google provides two key pages within your account: Search Services History and Search Services Personalization. The following steps show you how to disable both on the Google app or via the web.

Disable media saving in Search Services History

• Open the Google app on your mobile device.
• Tap your profile picture in the top-right corner.
• Select Search history.
• Make sure the Save media toggle is off. If it's blue (on), tap to turn it off.
• Next, tap the top entry labeled Search Services History. If it shows On, tap it and choose to turn the service off.

You can also access these settings from a desktop by visiting myactivity.google.com. Click on Search Services History and uncheck the box next to Save media. I recommend keeping both checkboxes unchecked to prevent any automatic saving of future media.

Turn off Search Services Personalization

• Go back to the Google app and tap your profile picture.
• Select Search personalization.
• Tap the top entry and ensure the service is set to Off.

Disabling personalization means Google will not use your search activity to tailor results, but it also prevents them from leveraging that data for AI training. These two settings are independent, so you should adjust both for maximum protection.

Optional: Auto-delete saved data after a specific period

If you do not wish to fully disable the features, you can set a time limit on data retention. Google offers three options: 3 months, 18 months, or 36 months. After the selected period, all saved media in your Search Services History will be automatically deleted. To configure this, go to the same Search Services History page and choose Auto-delete. I advise setting it to 3 months to minimize exposure.

Additional privacy steps

Finally, navigate to myadcenter.google.com and turn off Personalized Ads. While this does not directly affect AI training, it reduces the amount of data Google uses to profile you for advertising purposes. Combined with the previous changes, you can significantly shrink your digital footprint.

Are there any benefits to leaving it on?

Some users prefer keeping these features enabled because they enjoy personalized search results or find targeted ads convenient for shopping. Google's AI also becomes more accurate when trained on diverse media, potentially improving services like Google Lens, image translation, and voice recognition. If you trust the company to handle your data responsibly and value a more seamless experience, you might choose to stay opted in. However, the trade-off is a substantial loss of privacy.

Personally, I consider privacy paramount. That is why I run all my AI tasks locally using open-source models like Llama or Mistral, which never share data with any cloud provider. While not everyone can adopt that approach, disabling Google's media training is a simple and effective first step.

Historical context: Google's evolving data policies

This is not the first time Google has expanded data use for AI. In 2023, the company updated its privacy policy to allow scraping of public content from the web for training purposes. Earlier this year, it began using Google Docs and Sheets data to refine its Gemini model. Each change has tightened the grip on user information while offering a narrow opt-out window. The current update extends the same logic to multimedia content, reflecting Google's insatiable appetite for training data in the race to dominate the AI landscape.

Competitors like Apple and Microsoft have taken different approaches. Apple has committed to on-device AI processing for many tasks, while Microsoft offers users the ability to delete their training data from Copilot. Google, by contrast, relies on cloud-based AI and benefits directly from the scale of data collected through its ubiquitous search engine, Android ecosystem, and YouTube platform. As of 2026, Google handles over 8.5 billion searches per day, making it the largest single source of potential training material on earth.

The key takeaway is that users must stay vigilant. Privacy is not a default in the digital age; it requires active management. By following the steps outlined above, you can ensure that your images, voice, and videos remain yours alone. Remember to check your settings periodically, as Google may change them again without prominent notice.

Source:ZDNET News