Android SafetyCore has drawn renewed attention after users reported that the system component reappeared on their devices following updates. Google says the service supports on-device sensitive-content warnings, but a new developer API has raised fresh questions about how similar content classification could be used by other apps.
What Android SafetyCore does
Android SafetyCore provides shared infrastructure for safety features on Android devices. In Google Messages, it supports Sensitive Content Warnings that can blur potentially nude images and show a warning before users view, send or forward them.
Google says the classification happens exclusively on the device. The company says the feature does not send detected images, classification results or identifiable information to its servers.
For adults, Sensitive Content Warnings are opt-in. The setting is available in Google Messages. However, the feature is enabled by default for some supervised teen accounts.
Why users remain concerned
Many users were surprised to find Android SafetyCore installed without a clear explanation of its purpose. The component has no prominent app icon, which has added to confusion and concern.
Some users have also reported that the service reappeared after they removed or disabled it. This can happen when Google Play system components update, although Google has not described every reported installation case.
Android SafetyCore does not scan all photos on a device by default. Google says the image classification feature runs only when eligible content passes through Google Messages and Sensitive Content Warnings are active.
New API could extend content classification
Google’s developer documentation now references a ContentSafetyManager API in development. It would allow approved Android apps to request on-device content classification and receive a result.
The API can process image and other media data. It returns broad categories such as allowed, warning, blocked or unclassified.
The documentation does not state that every third-party app will receive access. It also does not confirm that the API directly uses the current SafetyCore component.
Still, the proposed framework could give developers a standardised way to classify content locally. Apps could then choose whether to display, blur or restrict the relevant media.
Privacy risks depend on app permissions
On-device classification does not give an app automatic access to a user’s photos or files. An app still needs the required permissions before it can submit media for analysis.
However, privacy advocates worry that a malicious app with broad media access could use the system to categorise sensitive content quickly. The classification feature alone would not send images away, but an abusive app could potentially collect or exfiltrate files through other means.
Google says Android SafetyCore is privacy-preserving and designed to protect users from unwanted content. The upcoming API, however, makes clear why transparency, permissions and strict access controls will matter as on-device safety tools expand.
0 responses to “Android SafetyCore API Raises New Privacy Questions”