Whenever a celebrity breaks up with their significant other, there's a running joke among normal people that this is their moment to shoot their shot. Until now, it has remained just that—a joke. But what happens when a person is able to take things a step further by enabling a machine learning algorithm, borrowing from easily accessible photos and videos, and utilizing open source code to put that person of your dreams in a sexual situation?
With the help of an open-source machine learning tool similar to TensorFlow, a Redditor named "deepfakes" has been able to create a new type of porn video by putting the faces of celebrities such as Scarlett Johansson, Taylor Swift, and Gal Gadot on adult film actresses. The faces are obtained through a combination of Google image search, stock photos, and YouTube videos.
"I just found a clever way to do face-swap," deepflakes told Vice, referring to his algorithm. "With hundreds of face images, I can easily generate millions of distorted images to train the network. After that if I feed the network someone else's face, the network will think it's just another distorted image and try to make it look like the training face."
While there are still some flaws to deepflakes's methods, it's believable enough to make you feel like there's something very wrong about these videos. And you wouldn't be wrong. "One important thing that always needs to happen is consent," porn performer Grace Evangeline said. "Consent in private life as well as consent on film. Creating fake sex scenes of celebrities takes away their consent. It’s wrong."
However, deepflakes doesn't necessarily see this matter as being that black and white. "Every technology can be used with bad motivations, and it's impossible to stop that,” he said, comparing his work to that of the recreation of Paul Walker in Furious 7. "The main difference is how easy [it is] to do that by everyone. I don't think it's a bad thing for more average people [to] engage in machine learning research."