Google is developing a new public Android API that will enable third-party applications to scan images and other files stored on the device and classify them according to their safety level.
The mechanism is built around the existing SafetyCore system component, which already blurs intimate images inside Google Messages. The new ContentSafetyManager will accept images, raw files, and other multimedia types for analysis.
After processing, the system will assign one of four statuses to the content: allowed, requires warning, blocked, or unclassified. The calling application will then decide whether to display the item, blur it, or hide it completely.
All analysis is designed to run locally on the device without sending any photographs or files to Google servers. The API itself does not grant access to the entire photo gallery; separate permissions must still be requested by the application.
Security researchers have raised concerns that malicious applications granted excessive permissions could route large volumes of user files through the classifier, construct detailed user profiles, or repurpose the results for unintended purposes. Google advises developers to request only the minimum necessary permissions.
SafetyCore previously drew criticism after being installed automatically without explicit user consent and without displaying a standard app icon. The component has been downloaded more than one billion times, prompting some Android users to install dedicated blockers that prevent the Play Store from reinstalling it after removal.
Google maintains that processing remains local and privacy-preserving, yet users continue to question why the component appeared on devices before any detailed explanations were provided.
