AI is a real problem on social media these days. As the abilities of generative software become increasingly more advanced, it’s becoming much more difficult to tell whether the image or video is “real” or not.
Nefarious content producers are swamping online channels with fake content so that we become less able to tell the difference, so that they can then begin slipping fake news videos and pleas for money amongst what we see every day.
There is another interesting approach: taking a copyrighted photograph and modifying it with AI so that it no longer breaks the terms of use. Look at the examples above: the original photograph (the man has a blue sleeve) shows the unnaturalness of the real world (a watch, legible text on a badge, and a random person’s foot behind the horse). The fake (the man has a green jumper to match the wall colour) shows an “perfect” version, in which the textures are unnaturally smooth, the hair on the horse’s mane is almost painterly, and the random foot and unphotogenically positioned chair have all been “fixed”.
Here’s an interesting video produced by Corridor Crew, in which the team teaches family members about how to spot fake content.
Leave a Reply