Google Uses Search Media for AI Training, Opt-Out Available in Settings

In early July 2026, Google introduced a policy to use media uploaded by users to its search services for AI training. According to an Engadget report citing TechCrunch, the change was implemented quietly and covers a broad range of data, including “images, files, audio, and video recordings.”

Specifically, when a user uploads a photo to Google Lens for visual search, that image data could potentially be used for training. Audio data from Google voice searches and files uploaded to Google Translate are also covered. However, services other than search, such as personal Google Photos, are currently excluded.

The fact that all users are opted in by default has sparked significant debate. Generative AI requires vast amounts of data, and existing data on the internet is becoming insufficient, which appears to be the background for this move.

Steps to Change Settings

Users who wish to opt out can exclude their data from training by making the following settings changes.

First, go to Google’s “Search Services History” page and uncheck “Save Media.” Next, on the “Search Services Personalization” page, change the settings so that data saving is not performed. These two steps prevent uploaded media from being diverted for AI training.

Additionally, to disable the AI summary feature displayed in search results, you can type “-AI” before your query. However, it is important to note that these settings changes only apply to data that Google explicitly collects, and some modern advanced AI techniques can access data without explicit permission.

The Pros and Cons of Automatic Opt-In

The most controversial aspect of this change is that data collection begins without the user’s explicit consent. From a privacy protection perspective, there is strong opinion that a major platform like Google should implement a clearer consent acquisition process when expanding its data usage scope.

In particular, Google Lens and voice search are features used daily, and in most cases, users upload data without intending it to be used for AI training. Some experts point out that the shift from opt-in to opt-out may violate regulations such as the EU’s General Data Protection Regulation (GDPR).

On the other hand, Google needs large amounts of data to improve the quality of its AI models and maintain competitiveness, and the diverse media uploaded by users represents valuable training data. The company has long been collecting data from public sources and its own services, and this expansion is an extension of that practice.

Industry-Wide Trends

Google’s move symbolizes the current scramble among technology companies to secure AI training data. Text data on the internet has been nearly exhausted, and companies are focusing on securing multimodal data such as images, audio, and video.

For example, the time-series forecasting model TimesFM 2.5, released by Google Research in early 2026, is one result of leveraging diverse data sources. Meanwhile, as Google expands consumer services such as the customized ticket feature for Ticketmaster in Google Wallet, this data collection policy may alienate privacy-conscious users.

Other major tech companies like Meta, Microsoft, and OpenAI are also steering toward using user data from their own services for AI training, but there are differences in methodology and transparency between companies. It has been pointed out that disparities in regulation and user evaluation may emerge between companies that opt users in automatically like Google and those that adopt a more explicit consent process.

Measures Users Can Take

At present, the scope of data Google uses for AI training is limited to media uploaded to search services. However, the possibility of future expansion to other services cannot be ruled out.

Measures users can take to protect their data include, first, implementing the settings changes mentioned above. Additionally, it is important to regularly review data management across Google services and develop a habit of deleting or downloading data as needed.

Furthermore, if you do not want to provide data for AI training, consciously choosing the content and format of media uploaded to Google services is one option. However, if you seek complete data protection, you may need to consider alternative services.

Editorial Opinion

In the short term, Google’s policy change is likely to improve the performance of its AI models, but it may also lead to an exodus of privacy-conscious users. In particular, compliance with data protection regulations is expected to be questioned in Europe and some U.S. states, and we anticipate that regulatory investigations or lawsuits from user groups may emerge within three to six months. Although opting out via settings is possible, the fact that the vast majority of users are unaware of its existence is problematic.

From a long-term perspective, a watershed moment will be whether such implicit data collection becomes an industry standard or is curbed by stricter regulations. Over a one-to-three-year span, the growing demand for data driven by AI advancement and the need for personal data protection are in a trade-off relationship, and how to balance these becomes a societal issue. It seems urgent to build mechanisms where companies enhance transparency and users can proactively manage their data.

The editorial board believes that this change should serve as an opportunity to rethink the value and rights of user data.

References

• Google now uses your uploaded search media to train AI — Engadget, published July 6, 2026
• Google Research Releases Time-Series Forecasting Model TimesFM 2.5 — Related article on this site
• Google Wallet Custom Ticket Feature for Ticketmaster — Related article on this site