A new app that lets people substitute their faces with celebrities’, with uncanny accuracy, has led to major concerns about what else it is doing with its users’ likeness.
The Chinese app, called ZAO, is currently the most popular download in the iOSApp Store in China and has been downloaded millions of times, and the interest was such that the app nearly broke. Videos taken from the app have very quickly gone viral, with users sharing videos of themselves swapped into famous scenes from Game of Thrones or the films of Leonardo di Caprio.
But the huge success has led to immediate questions over what exactly the app is doing with people’s faces and how similar technology could be abused in the future.
One of the chief concerns is about the privacy of the app. When a user signs up to use it, they approve its terms and conditions and then upload photographs of their face.
That face is then used to swap into videos including sports or film stars. They can then be passed onto friends.
But those same images could be used forever by the app for whatever purposes the developers want, some have suggested.
Part of the privacy policy that users agree to makes clear that anyone uploading their images to the app surrender the intellectual property rights to that image. ZAO can therefore use personal images, which appear to have been uploaded only to have fun with the app, for its own marketing images.
ZAO responded to those concerns on Chinese social media site Weibo. But it did not make clear what exactly it would do, or how it was using those images at the moment.
“We thoroughly understand the anxiety people have towards privacy concerns,” the company said, in a statement reported by Reuters. “We have received the questions you have sent us. We will correct the areas we have not considered and require some time.”
The app has also led to concern around “deepfakes” – the technology that allows people to convincingly swap faces into other videos, or to use AI to generate videos of people saying things they have not said. AI experts have repeatedly warned that the technology seen in ZAO could be used to undermine trust, by creating false videos of real people saying or doing entirely fake things.
