Trust No One: Inside The World Of Deepfakes

In the halcyon, early days of smartphones, many people used face-swapping apps, largely for amusement, but the rapid advance of this technology, plus the feeding frenzy that is the internet, has transformed a seemingly harmless activity into something much more sinister, capable of spreading malicious misinformation via the far-reaching tentacles of social media. Can you ever trust what you see? It is a disturbing thought.

Deepfakes are fake videos, created using the power of artificial intelligence, in which a person in an existing image or video is replaced with someone else’s likeness, fooling people into believing that what they are seeing is authentic. The technology is cheap, easy to use and access, and it is difficult to bring a lawsuit against a deepfaker.

As with many internet trends, it began with pornography. Fake celebrity porn has become an entire internet subculture, with Emma Watson (340 existing videos) the most deepfaked celebrity in the world, closely followed by Scarlett Johansson (below). If you can make anyone say or do anything anywhere, the potential for political or criminal exploitation is immense: the ability of deepfake AI to synthesise voice and enable audio impersonation has already been used successfully for financial extortion and blackmail.

Michael Grothaus has written a readable, thought-provoking, if slightly repetitive warning from the internet’s underbelly. He over-focuses perhaps on the celebrity-porn aspects of deepfaking and deals only fleetingly with the more positive uses of the technology (for example, allowing people with eating disorders to view images of themselves in a healthier body), but he interviews some shady characters and raises interesting questions. He even commissions a deepfake of himself committing a crime, and another in which his father is brought back to life; he is, not surprisingly, uncomfortable with the results.

Be warned and be afraid.