Mark Zuckerberg, Barack Obama, Vladimir Putin and Nancy Pelosi. The list of politicians as well as other types of rulers, as depicted in manipulated videos - so-called deepfakes - has started to be long. Sometimes it is humor and satire, but the new AI technology can also be used for political purposes, to spread disinformation or dirty political opponents, as reported by Cultural News in the past.
But now deepfakes face resistance. This fall, researchers at Lund University will begin work on a new app that will be able to reveal fake or manipulated news material in Sweden, using their own AI algorithms. It reports the Science Radio in SR.
- We may get used to the fact that we cannot rely on an image or an audio file, says Kalle Åström, professor at the Mathematics Center at Lund University, to the Swedish Science Radio.Growing trend
Similar technologies are now being developed in different parts of the world, as the technology needs to be adapted to a specific language. The programs are primarily intended to be used by journalists, but should also be available to the public.
Technology is called GAN, where two networks "compete" against each other. One tries to create such realistic manipulated clips, while the other tries to reveal them. The researchers at Lund compare the technology with the fight against computer viruses, where the technology constantly needs to be developed.
- Viruses are also created all the time, but at the same time there are people who are fighting against virus production, says political scientist Anamaria Dutceac Segesten from Lund University, to SR.