Google’s Android System SafetyCore sparked a wave of backlash after quietly appearing on millions of Android phones through Play System Updates. Social media quickly filled with claims that it was secretly scanning users’ photos or acting as spyware.
After reviewing Google’s documentation, independent analysis from the privacy-focused GrapheneOS project, and reporting from major tech publications, the evidence paints a more nuanced picture. SafetyCore is a real content-classification service, but it isn’t secretly uploading your photo library to Google’s servers. The biggest controversy is how Google installed it with little warning.
What is Android System SafetyCore?
Android System SafetyCore (com.google.android.safetycore) is a Google system service available on Android 9 and newer devices. Rather than being a normal app, it provides shared machine-learning infrastructure that supported apps can use to classify content such as nudity, spam, scams, and malware.
Unlike the rumors suggested, SafetyCore doesn’t independently scan your phone. An app has to specifically request content classification through Google’s Content Safety API.
Does SafetyCore scan all your photos?
The short answer is no.
Current evidence shows SafetyCore only processes content when a supported app asks it to. Its primary public use today is Google Messages’ Sensitive Content Warnings, which can blur suspected nude images before you view or send them.
Google says the classification happens entirely on-device, meaning images aren’t uploaded during normal processing. GrapheneOS independently confirmed that the service doesn’t perform client-side scanning that reports image content back to Google or anyone else.
SafetyCore currently powers one of Google Messages’ most prominent privacy features.
When enabled, it can:
- Blur suspected nude images before they’re opened.
- Show a warning before viewing sensitive content.
- Warn users before sending or forwarding similar images.
For adults, the feature is optional and can be turned on or off in Messages settings. For supervised accounts under 18, it’s enabled by default as part of Google’s family safety features.
Why people got angry
The criticism surrounding SafetyCore isn’t just misinformation. Several complaints are legitimate.
Silent installation
SafetyCore arrived through Play System Updates without a clear notification or permission request. Many users only discovered it later because it doesn’t appear in the normal app drawer and requires enabling Show System Apps to find it.
It often comes back
Users frequently report that uninstalling or disabling SafetyCore isn’t permanent, as later system updates may reinstall it.
It’s closed-source
Unlike much of Android’s open-source framework, SafetyCore itself isn’t open-source. That lack of transparency has made privacy-focused users skeptical, even if the current functionality appears limited.
Performance concerns
Some users also reported battery drain, increased memory usage, or slower performance around the time SafetyCore appeared. While those reports aren’t conclusively linked to the service itself, they became part of the broader backlash.
Can Google expand it later?
Google has introduced APIs that allow developers to use SafetyCore’s on-device classification capabilities in their own apps. That has raised concerns about potential future expansion.
Privacy advocates worry that more apps could begin requesting content classification or that Google’s use cases could broaden over time. However, there is currently no credible public evidence that SafetyCore secretly uploads classified image data or continuously scans every photo stored on Android devices.
Can you remove SafetyCore?
On many phones, yes.
Go to:
Settings → Apps → Show System Apps → Android System SafetyCore
From there, you can choose Uninstall or Disable, depending on your device. Removing it doesn’t break Android’s core functionality, but features like Google Messages’ Sensitive Content Warnings will no longer work.
The bottom line
Android System SafetyCore isn’t the always-on spyware that many viral posts claimed it was. Based on the available evidence, it performs on-device, app-triggered content classification rather than constantly scanning your entire phone.
At the same time, users’ frustration is understandable. Google’s decision to silently install a hidden system component, provide only a vague description, and allow it to reappear after updates created a trust problem that continues to fuel the controversy—even if the technology itself appears far more limited than the internet initially believed.
