Consider this a belated public service announcement: if you use any Google service, you may be unknowingly contributing to the training of its artificial intelligence models. A quietly introduced update to Google’s privacy settings now enables the company to store more of your personal data—specifically media such as images, files, audio, and video recordings—and use that data to improve its AI systems. The change, which took effect in June 2026, was communicated to users via email but otherwise flew under the radar. Essentially, Google opted everyone into this expanded AI training under the guise of offering more control over saved history and personalized recommendations.
The core issue revolves around two new settings: Search Services History and Personalized Recommendations. These settings allow Google to configure how your activity is used to tailor your experience and how long your web and app activity is retained. The problem is that both settings are enabled by default, meaning every Google user who does not take proactive steps is now supplying the company with valuable training data. This includes data from Google Search, Maps, Shopping, Flights, Hotels, Translate, and News. For instance, snapping a photo using Google Lens to search for a product or landmark will now likely save that image for AI training. Similarly, using the newer Search Live feature for voice input in the Google app, or any voice search via Google, will store those audio recordings. Even practicing pronunciation on Google Translate can result in your audio being saved and analyzed.
Google’s own help documentation confirms this dual purpose. The company states, “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.” It further notes that human reviewers may access this data to help protect users and the public. While some of this storage is temporary and necessary for the product to function, Google explicitly states that saved media can be retained specifically for AI training. This marks a significant shift from earlier practices, where data retention was primarily for personalization and analytics.
This move aligns with a broader industry trend. Rather than relying solely on web-scraped data, tech giants are increasingly exploiting user-generated content to train their AI. Meta, for example, trains its AI on images and media uploaded by users, as well as footage from its Ray-Ban smart glasses. Google’s approach, however, is more pervasive because of the sheer number of services it offers and the default opt-in nature of the settings. The company has long used search history, location, and browsing data to personalize ads and recommendations, but now the scope has expanded to include rich media that can teach AI systems about visual and auditory patterns.
Understanding the new settings and how to take control
The good news is that users have options to protect their data. First, you can adjust your preferences on the Search Services History page and the Search Services Personalization page. On the Search Services History page, you will see two checkboxes: one for “Search Services History” and one for “Save Media.” By unchecking “Save Media,” you prevent Google from storing future images, audio, and video content for AI training, while still allowing other activity history to be saved for personalization. You can also uncheck both boxes to disable all history tracking for search services. Additionally, you can set an automatic deletion schedule for saved data—options include three months, 18 months, or 36 months. This is a good way to limit the duration that your data remains available for training.
It is important to note that the new Search settings are separate from the older “Web & App Activity” settings. Previously, controlling search history was part of Web & App Activity. Now, with the June update, Google split them into two distinct controls. This means that if you previously disabled or limited Web & App Activity, it no longer affects your Search Services History. You must explicitly manage the new Search settings to stop the collection of media for AI training. Similarly, adjusting YouTube History or Location History requires separate steps.
From the Search Services History page, you can navigate to a broader privacy settings hub where you can review and modify settings for Web & App Activity, Timeline, YouTube History, and more. Google provides a centralized dashboard, but because the changes are scattered across multiple pages, it is easy to miss the new defaults. The company also sent an email to users in June explaining the update, but many people either ignored it or deleted it without reading. As a result, the vast majority of Google users are now contributing to AI training without explicit consent.
For those who want to go further, consider deleting existing saved media. The Search Services History page allows you to manually delete all stored media or specific items. You can also pause all future collection by toggling the “Search Services History” switch to off. This will stop Google from saving any activity related to search services, including media. However, remember that some media may still be temporarily stored for technical reasons (e.g., to process a translation request), but Google claims it will not be retained for AI training if you have opted out.
Implications for privacy and the AI arms race
This development highlights an ongoing tension between user privacy and the insatiable appetite for data by AI companies. Generative AI models, like Google’s Gemini and others, require vast amounts of diverse data to improve accuracy, reduce bias, and learn new tasks. User-generated content—photos, voice recordings, search queries—is highly valuable because it reflects real-world human interaction. However, the practice of default opt-in and buried settings raises ethical concerns. Critics argue that such changes violate the spirit of GDPR and similar privacy laws in regions like Europe, where consent must be freely given and specific. Google may face legal challenges as a result.
On a practical level, users concerned about their digital footprint should take immediate action to review and adjust these settings. The steps are not difficult, but they require a few minutes of effort. Beyond the emotional cost of losing control over one’s data, there are tangible risks: sensitive images from Google Lens, private voice conversations, and even audio of children could end up being used to train AI—and possibly retained indefinitely unless deletion settings are configured. Google has stated that it uses “human reviewers” as part of its safety measures, which means actual people may view your data during the training process. This is a sobering reality for anyone who values confidentiality.
The broader context is that we are witnessing a race among the largest tech companies to amass as much data as possible. OpenAI, Meta, Microsoft, and Google are all competing to build the most powerful AI. The line between improving user experience and surveillance is blurring. For Google, the ability to train AI on user media could accelerate the development of more capable visual and voice assistants, potentially resulting in better products. But for users, the cost is a further erosion of privacy. The company’s help text acknowledges that saved media “is also used to develop and improve Google services and technologies, including AI models.” This is a clear admission that your media is a resource for machine learning, not merely a tool for personalization.
To summarize the key steps: Go to the Search Services History page (accessible via your Google Account settings). Uncheck the box next to “Save Media” or turn off Search Services History entirely. Set an auto-delete timer to remove older data. Also visit the Web & App Activity page and consider limiting or pausing that history, though note that it no longer covers search services. Finally, periodically check your privacy dashboard for any new defaults that may appear as Google continues to update its policies. While these actions will not erase all data previously collected, they will stop the future flow of your media into Google's AI training pipeline.
Source: TechCrunch News