Twitter starts to tackle doctored videos—not a moment too soon

TRAWLING THROUGH oceans of spam, hate speech and violent pornography, as social-media content moderators must do, is a dismal and unrelenting task. The human filters at Twitter will soon have to catch even more grot. On March 5th the social-media platform started alerting users to fraudulent videos, audio and images that have been altered or fabricated, material the company calls “synthetic or manipulated media”.

Such bogus content has proliferated in recent years. The most sophisticated forgeries, known as “deepfakes”, use artificial intelligence to edit someone’s face into a film. One well-known example that went viral in 2018 put the words of Jordan Peele, a comedy actor, into Barack Obama’s mouth, as he jokingly badmouthed Republican rivals. Other fraudulent videos rely on simpler methods. In May 2019 Twitter users circulated a video of Nancy Pelosi, the Democratic Speaker of the House of Representatives, which had been slowed down to make it appear that she was drunkenly slurring her speech.