Google introduces private computing services for Android 12 Beta

In a recent blog post, Suzanne Frey, Google’s vice president of security and privacy products for Android and Play, is pleased to announce a private computing service to the majority of Android users. It is reported that Private Compute Core aims to provide an open-source and secure environment, isolated from other operating systems/applications, and to build a privacy protection bridge between user devices and the cloud.

In each new version of Android, Google will strive to bring more privacy protection features of Private Compute Core. Today, we have seen improvements in three areas including Live Caption, Now Playing and Smart Replay.

  • First of all, Live Caption will be able to use the voice recognition function of the local device to add subtitle annotations to any media;
  • Secondly, the Now Playing function can identify the music being played nearby and display relevant information (such as song/artist name) on the device lock screen;
  • Then, Smart Replay can update the content in the message application and give smart reply content suggestions.

With the support of the private core computing function, users can not only ensure the privacy of the information stored on the device but also allow the device to call cloud functions (such as downloading new song classification/voice recognition models) without paying for privacy.

Google said that Android can ensure that sensitive data processed in Private Compute Core will not be shared with any application without the user taking any measures. For example, before the user clicks on Smart Recovery, the operating system will isolate your reply content from the input method/current App hiding. As for the private computing service itself, Google said that they will continue to update related machine learning models, and then obtain these updates through private channels.

The Android operating system will prevent any function in Compute Core from directly accessing the network, but can only communicate with private computing services through a set of targeted open-source APIs. In addition, private computing services will strip off information that can be used to identify users, as well as technologies such as joint learning, joint analysis, and private information retrieval.

Finally, Google promised to publicly release the source code of its private computing service for review by security researchers and teams other than Google.

Leave a Comment