Google Search uploads can now train AI unless users manually opt out
Your Google Lens images and voice queries may become AI training data by default.
Google has quietly updated its Search services privacy settings so that any media uploaded during interactions with products like Google Lens, Search Live, Translate, or voice search may be saved and used to improve its AI models. The change, rolling out over the next few months, separates the new "Save Media" sub-setting from the existing Web & App Activity control. Users who do not opt out will have their images, files, audio, and video automatically stored in Search Services History and potentially fed into Google's AI training pipeline.
For enterprise users, the risk is acute. A single work screenshot taken with Google Lens, a translated confidential document, or a voice query containing sensitive information could become part of Google's model training data. The opt-out is manual and does not affect previously saved media, which continues to be used unless explicitly deleted. Moreover, other Google products like Gemini Apps, NotebookLM, and YouTube have separate controls. This fragmentation creates a compliance headache for organizations that rely on Google Workspace or allow employees to use consumer search tools for quick problem-solving.
- Google's new 'Save Media' setting automatically stores images, audio, and video from Search, Maps, Translate, and Lens for AI training unless manually disabled.
- Opting out does not delete previously saved media, which may still be used to improve Google's AI models.
- Businesses face data governance risks as routine work uploads (screenshots, translated docs, voice queries) could become training material without clear user consent.
Why It Matters
Employees' routine search actions may feed sensitive corporate data into Google's AI training pipeline.