Google expands AI training to include media uploaded through Search tools
Google now uses some media uploaded through Search tools to train its AI, and users must opt out via account settings.
Google has updated the way it collects user data for artificial intelligence development, expanding its policies to include media uploaded through several of its Search-related services. The change, first reported by TechCrunch, means that certain images, audio, video and files submitted through Google Search tools may now be used to help improve the company’s AI models.
Table Of Content
The update applies to content uploaded directly through Search features, rather than all Google services. While the change has attracted attention from privacy advocates, users who prefer not to have their uploaded media used for AI training can adjust their account settings to limit data collection.
Search uploads now fall under the AI training policy
According to the updated policy, Google may use media uploaded through Search-based services to develop and improve its artificial intelligence technologies. The material includes images, files, audio recordings and video recordings submitted while using supported Search features.
In practical terms, this affects media uploaded via tools such as Google Lens, where users submit photographs to identify objects or gather visual information. It also covers voice recordings made during Google voice searches and content uploaded through Google Translate when using its translation features. The policy applies specifically to Search-related services rather than to all Google products.
The change does not extend to personal media stored in Google Photos. Content saved in that service is not included under this particular Search policy, providing a distinction between files intentionally uploaded for Search functions and those stored for personal use.
The revised approach reflects the growing demand for data needed to train increasingly sophisticated generative AI systems. Large language models and multimodal AI platforms require substantial volumes of text, images, audio and video to improve their accuracy and capabilities. As technology companies continue expanding their AI offerings, user-generated content has become an increasingly valuable resource for model development.
Although Google has not widely promoted the policy update, the revised terms mean users are automatically included unless they manually change their account settings. This opt-out approach has raised fresh questions about transparency and how technology companies communicate changes involving personal data.
Users can adjust settings to reduce data collection
Users who do not want uploaded Search media to contribute to AI training can change their Google account settings.
One step involves visiting the Search Services History page and disabling the option labelled “Save Media”. Users can also visit the Search Services Personalisation page to ensure that Search activity and uploaded media are not retained for personalisation. Adjusting these settings limits the amount of Search-related media that Google stores and may reduce its availability for AI development.
While these settings provide users with greater control over their data, they require manual action. Anyone who does not change the default configuration remains enrolled under the updated policy.
The change highlights a broader trend across the technology industry, where companies increasingly rely on user interactions to refine AI systems. Images uploaded to identify plants, products, or landmarks, voice recordings made during spoken searches, and files submitted for translation are valuable real-world examples that can improve machine learning performance.
Privacy experts have long argued that clearer communication is needed whenever companies broaden the ways personal information may be used. Automatic enrolment, even with opt-out controls, remains a point of debate as AI becomes more deeply integrated into consumer services.
AI data practices remain under growing scrutiny
The latest update comes as artificial intelligence companies face increasing scrutiny over how they obtain and use training data. Questions surrounding consent, transparency and copyright have become central issues as generative AI tools become more widely adopted.
Technology firms have defended the need for extensive datasets, arguing that diverse, high-quality data enable AI systems to produce more accurate and useful results. At the same time, creators, publishers and privacy advocates have raised concerns about whether individuals fully understand how their content may be reused once it is uploaded to online platforms.
Google’s revised Search policy is another example of how AI development continues to reshape longstanding data-collection practices. While the company provides options for users to manage certain settings, the responsibility currently rests with individuals to review their preferences and decide whether they want their Search uploads to contribute to future AI improvements.
The issue extends beyond Google, with several AI developers facing criticism for collecting publicly available or user-submitted content for training. Debates involving artists, musicians, writers and other creators have highlighted the challenges of balancing rapid AI innovation with expectations around privacy, ownership and informed consent.
As AI capabilities continue to evolve, policies governing how user-generated content is collected and processed are likely to remain under close examination by regulators, consumer groups and the wider technology industry.
