“We are crossing over into an era where we have to be skeptical of what we see on video,” says John Villasenor, a senior fellow at the Brookings Institution. Villasenor is talking about deepfakes—videos that are digitally manipulated in imperceptible ways, often using a machine-learning technique that superimposes existing images or audio onto source material. The technology’s verisimilitude is alarming, Villasenor argues, because it undermines our perception of truth and could have disastrous consequences for the upcoming U.S. presidential election.
“I do think deepfakes are going to be a feature of the 2020 elections in some way,” Villasenor says. “And their shadow will be long.”