Deepfakes could pose a greater threat than the fake news and Photoshopped memes that littered the 2016 presidential election because they can be hard to spot and because people are -- for now -- inclined to believe that video is real. But it's not just about individual videos that will spread misinformation: it's also the possibility that videos like these will convince people that they simply can't trust anything they read, hear or see unless it supports the opinions they already hold. Experts say fake videos that will be all but impossible to identify as such are as little as 12 months away.
Jonathon Morgan, the CEO of New Knowledge, which helps companies fight misinformation campaigns and has done some analysis for CNN, sees troll farms using AI to create and deliver fake videos tailored to social media users' specific biases. That's exactly what the Russian-backed trolls at the Internet Research Agency did during the last presidential election, but without the added punch of faked video.