While Android 12 brings along a bunch of new privacy-focused features such as privacy dashboard and privacy indicators, one feature that often gets overlooked is the Private Compute Core. Private Compute Core is part of Google’s vision to help ensure privacy while handling sensitive data on your device. In this article, we will take a look at what Private Compute Core is all about and its capabilities.
What is Android Private Compute Core (2021)
We have discussed the basics of Android Private Compute Core, along with the features supported in Android 12. There’s also a section to enable one of the smart features in Android 12, which you can navigate to using the table below:
What is Android 12 Private Compute Core?
Android 12’s Private Compute Core is a unified secure environment that acts as a sandbox for features that process sensitive data. With the help of Private Compute Core, Android isolates ML-enabled sensitive data from the rest of the OS and other apps you may have installed on your device. Thanks to this approach, features inside Private Compute Core do not have direct access to the Internet.
Having said that, the scope of Private Compute Core is noticeably limited at the moment. However, Google has promised to add more features to Private Compute Core with each new Android release.
What is Android Private Compute Services?
Machine learning models are often updated for enhancing overall efficiency and accuracy. Hence, there has to be a way for Google to push the updated ML models to the device in a secure manner that preserves privacy. To bridge the gap between your device and the cloud, Google has introduced Android Private Compute Services.
Private Compute Services offers a set of open-source APIs that lets components in Private Compute Core access the network. As per Google, these APIs will strip out identifiable information and utilizes privacy-focused techniques such as Federated Learning, Federated Analytics, and Private information retrieval.
Private Compute Core: Supported Features in Android 12
With that said, let’s take a look at the features supported by the Private Compute Core:
- Live Caption
Live Caption automatically generates captions to media you play on your phone. The feature uses Google’s on-device speech recognition to generate captions. Considering the fact that it requires access to the media for the feature to work as intended, Google has made the right call to integrate it to Private Compute Core.
- Now Playing
Now Playing is a Pixel-exclusive feature that automatically identifies music playing near you without having to use a separate app to identify songs. It accesses the microphone, and Private Compute Core will help isolate the data processing involved here.
- Smart Reply
Smart Reply is another crucial area where Private Compute Core is beneficial. Since the feature relies on suggesting relevant replies to your texts in messaging apps, privacy is a core aspect here. According to Google, Android will keep your reply hidden from the keyboard and the messaging app until you choose to tap the suggestion.
Check Private Compute Core Settings
- Open the Settings app and tap on Privacy. Under Privacy Settings, you will see a new option called ‘Private Compute Core’. Do note that it will appear as ‘Device Personalization Services’ on Android 11 phones or older.
2. Here, you will see the option to allow or disable smart text suggestions on the keyboard and clear data. To disable keyboard suggestions, tap on Keyboard and turn off the toggle.
3. Similarly, you can clear all the data stored so far. The available options are last hour, last 24 hours, and all time. You could also enable the ‘Also clear my Now Playing history‘ checkbox if you would like to remove that too. Once done, press ‘Clear data’.
Android Private Compute Core Explained in Detail
Private Compute Core indicates that Google is betting big on ML features for Android going forward. It will be interesting to see the kind of features that will make their way to the Private Compute Core umbrella in future iterations of Android. And thanks to its open-source nature, security researchers can audit the Private Compute Core to ensure the feature’s safety. What other features would you like to see in Private Compute Core? Share your thoughts in the comments.
