Columnists
Entering a World of Deepfaking
The New York Times reported a few weeks ago on a couple of potentially dangerous developments in the world of deepfaking—using artificial intelligence and machine learning to produce images that can’t be distinguished from originals, and manipulating old images to give them a digital afferlife.
The first development involved a Tom Cruise impersonator, who worked with a visual artist to produce “startlingly accurate” videos of what appeared to be the actor himself. I don’t know which is more alarming—the creation of a ‘fooled me’ video, or the fact that Tom Cruise is attracting impersonators and trying to vault himself into the same league as Elvis Presley, Liza Minelli and Cher. He certainly didn’t have my permission to get himself impersonated.
The second development was the animation of old photographs—more than 26 million of them—by a program called Deep Nostalgia. The program takes a still photograph, digitizes it, and has the subjects moving about and smiling. Now that’s eerie. When I see a photograph of Abraham Lincoln, I think of him as well and truly deceased; his smiling and moving for me at the whim of some punk software developer would only cheapen him—and unnerve me.
The developer of the Tom Cruise video says he spent two months training his computer to model Mr. Cruise’s facial expressions and that it’s “not something you can do at a home computer.” The Deep Nostalgia developer assures us that his technology is safe because his company only works with other companies that he trusts not to abuse it. All the same, the pace of innovation is such that the technology could land up on home computers—or smartphones—within a year or two; and if there is a demand for the technology, it will develop no matter who originally trusted whom not to do so.
The benign future uses of the technology have been called “endless.” Those listed by its proponents include the aging or de-aging of actors in old movies, improving the dubbing of voices on videos, and helping corporate executives speak more warmly to foreign audiences.
I can see the possibilities for personal use of the technology. While I would not call myself unduly vain, I could add some spark to my house by adorning my mantlepiece with pictures of me standing beside Desmond Tutu when he won the Nobel prize in 1983, or performing on stage for the Royal family with the Beatles in 1963.
But there is one potential use of the technology that blackens the sunny sky: the compromising photograph. What is to stop my enemies from putting together a fake photograph of me at, say, a Conservative party rally? I’d have to cough up dozens of dollars to keep it out of circulation. If this technology becomes widely available, the incidence of blackmail will surely increase. For blackmailers, it would be like shooting at a barn door.
There is also—fortunately for me—the potential for a rebound. If a copy is indistinguishable from the original, how can anyone prove it really is genuine, short of corroborating the photo evidence by seeing me there personally? Taking my cue from Donald Trump, I could denounce the photo as a fake. And a judge would likely believe me. Therefore, courts won’t be able to rely on photographic evidence anymore. Acquittals will become the norm. Criminals will roam the streets, Anarchy will ensue.
It’s hard to predict what kind of day to day life deepfakery will take us into. One thinker quoted in the article imagines a future in which “whole aspects of our personalities could be simulated after our death, trained by our voices on social media.” I think I’d choose to go through door number two myself—not knowing what lay behind it— rather than this door number one.
Whatever the future holds, I have one request. Please don’t perpetuate me as a Tom Cruise impersonator. Or, for that matter, as an Abraham Lincoln.
Comments (0)