Deepfakes: how artificial intelligence is being prostituted to create porn

Humans proclivity for sensual visuals has driven tech forward for centuries. The camera wouldn't have made it off the ground had it not been put to use capturing scantily clothed models for "artist drawings".

Our nostalgia for the VHS would not be what it is were it not for the fact that, due to its cheap, quick and dirty nature it became the staple for porn "straight-to-video" movies. And even the internet would not have become the integral part of our lives it is today if it were not for distribution of illicit material.

Thanks to someone's desire to see Gal Gadot naked, AI has become the bastion for digital facial manipulation.

In December 2017 a Reddit user called "deepfakes" decided to create a blue movie featuring the Wonder Woman actor by creating software that could superimpose her face seamlessly onto the body of a porn actor. The realism he achieved made the video and the tech behind it a sensation.

He uploaded the tech for free and though Reddit would go on to ban the subreddit due to a violation of its content policy - "specifically our violation against involuntary pornography" - it was too late. There are now many sites dedicated to porn with celebrities "deepfaked".

The integration of AI would take this practice to a new level. Last year in August at computer graphics conference Siggraph it was revealed that the technology has evolved to allow a source actor to be able to control a realistic virtual avatar to create human portrait videos in real time.

Furthermore it was revealed that companies have started feeding an AI bot photographs to teach it how the human face moves. The result of which allows a person to take a bevy of static images of someone's face and build a 3D rendering that can be "puppeted" to perform any realistic facial movement.

In other words, anyone who enjoys taking selfies - famous or not - can become a target.

Coupled with the rise of voice replication software the potential to create an international incident is high, to the point where the Defense Advanced Research Projects Agency of the US has set aside $28m to fund new tech that can identify and debunk manipulated videos.

At this time experts are able to debunk videos because deepfakes don't blink … yet. What will happen when alternative facts speak for themselves?