Fake America Great Again: Inside the Race to Catch Real Fakes Using AI

Photo fakery is far from new, but artificial intelligence will completely change the game. Until recently only a big-budget movie studio could carry out a video face-swap, and it would probably have cost millions of dollars. Artificial Intelligence (AI) now makes it possible for anyone with a decent computer and a few hours to spare to do the same thing. Further machine-learning advances will make even more complex deception possible—and make fakery harder to spot. These advances threaten to further blur the line between truth and fiction in politics. Already the internet accelerates and reinforces the dissemination of disinformation through fake social-media accounts. “Alternative facts” and conspiracy theories are common and widely believed. Fake news stories, aside from their possible influence on the last US presidential election, have sparked ethnic violence in Myanmar and Sri Lanka over the past year. Now imagine throwing new kinds of real-looking fake videos into the mix: politicians mouthing nonsense or ethnic insults, or getting caught behaving inappropriately on video—except it never really happened.

Perhaps the greatest risk with this new technology is not that it will be misused by state hackers, political saboteurs, or Anonymous, but that it will further undermine truth and objectivity itself. If you can’t tell a fake from reality, then it becomes easy to question the authenticity of anything. This already serves as a way for politicians to evade accountability. President Donald Trump has turned the idea of fake news upside down by using the term to attack any media reports that criticize his administration. He has also suggested that an incriminating clip of him denigrating women, released during the 2016 campaign, might have been digitally forged.