Ars Technica: Deepfakes are getting better—but they’re still easy to spot

Concerns about malicious use of those advances have given rise to a debate about whether deepfakes could be used to undermine democracy. The concern is that a cleverly crafted deepfake of a public figure, perhaps imitating a grainy cell phone video so that it’s imperfections are overlooked, and timed for the right moment, could shape a lot of opinions. That’s sparked an arms race to automate ways of detecting them ahead of the 2020 elections. The Pentagon’s Darpa has spent tens of millions on a media forensics research program, and several startups are angling to become arbiters of truth as the campaign gets underway. In Congress, politicians have called for legislation banning their “malicious use.”