Wednesday, January 24, 2018

Google App Matches Your Face

A Google app that matches your face to artwork is wildly popular. It’s also raising privacy concerns.

By Hamza Shaban

A Google app that matches people's selfies to famous works of art and encourages users to share the side-by-sides on social media leaped to the top spot on the iTunes App Store charts this weekend, ahead of YouTube, Instagram and Facebook's Messenger, but it has also drawn concerns from some that the privacy of the users may be at risk.


The latest version of the Google Arts & Culture app allows users to match their selfies against celebrated portraits pulled from more than 1,200 museums in more than 70 countries. The find-your-art-lookalike feature has been available since mid-December, but the app has rocketed to viral status as more users shared their matches on Facebook, Twitter and Instagram over the holiday weekend, in a mix of implausible, absurd and “spot-on” comparisons. People have also tested out the app using their dogs and pictures of celebrities and President Trump.

But not everyone was willing to snap away. Some people expressed skepticism over the privacy of the facial information users have been sending to Google.

The app works by using machine learning to recognize a person's face in the selfie, including the position of the head. It then compares the face to a bank of selected artwork to find matches.

Google says that the selfies are not being used to train machine learning programs, build a database of faces or for any other purpose. “Google is not using these selfies for anything other than art matches,” said Patrick Lenihan, a company spokesman.

The Arts & Culture app also says in one of its prompts that Google “will only store your photo for the time it takes to search for matches.”

The Arts & Culture app is one of the latest examples of how tech companies are implementing facial recognition technology. Google already uses it in its Photos service, which 500 million people use every month. Photos sorts pictures by people, places and things, and includes a feature that nudges users to share photos they have taken of their contacts, which the service recognizes.

In another example of the development of Google's image recognition, a feature was added to Photos in October that lets users sort pictures of their pets, even differentiating among dog breeds. In December, Facebook began flagging users that appeared on the social network without being tagged. Although that feature was designed to enhance users' privacy and control, it also highlighted how well Facebook's platform recognizes people's faces without much input from users. And in September, Apple's Face ID, introduced alongside its latest mobile phone, the iPhone X, sparked debate over the privacy and security of using a person's face to unlock the device and enable applications, including mobile payments.

No comments:

Post a Comment