Bogus images of angry dark-skinned men and bloodied blond women were quickly flagged as fakes, but the quality of the artificial intelligence is only bound to improve.
You have seen those AI-generated images circulating on the Internet in the last few weeks: Pope Francis in a puffy white parka, Emmanuel Macron in the mud, Donald Trump being forcibly arrested… Those images went viral. Such pranks, of course, might seem mostly harmless, and have been quickly flagged as fakes.
However, AI-generated images will no doubt be used for more dangerous purposes.
In Germany this week, we already saw it taken further than an innocent joke. Norbert Kleinwaechter, a deputy chairman from the far-right party Alles für Deutschland (AfD), recently posted on his Twitter account, several AI-generated images: one depicted a young blond woman, her face covered in blood, another showed a climate activist screaming.
Yet one particular image stands out, with an inscription — “No more refugees!”
In it, we see a group of angry and aggressive-looking, dark-haired bearded men, shouting. Kleinwaechter did not reveal that the images were not actual photographs and that the people were not real — though Twitter users rather quickly pointed out that they were bogus, noting certain details such as distorted faces in the background.
Point of view v. Reality
Kleinwaechter declared in a video that he wants to continue to use AI images, explaining that the migrant picture is, to his opinion, appropriate, even if it is a fake: “We have a hoard of migrants in the streets of Berlin, and they make the city unsafe and yell.”
Clearly, AI is helping them to spread their own perspective on immigration, as far of reality as they could be.
The AfD party also denies any attempt at disinformation, and explains to the daily Frankfurter Allgemeine Zeitung that those are “symbolic images” that they use from “image banks.” One AfD member said their movement is “much indebted” to this new form of technology, as it will help them to “generate images based on stereotypes.”
One of the fake AI-generated images AfD is spreading
AI keeps improving
Joachim Paul, an AfD deputy, declared in Freilicht, the blog of the far-right magazine “Identitarian Movement”, that he wants to use AI for the purpose of right-wing politics.
With the arrival of artificial intelligence, AfD has a tool to “make the world the way they present it: menacing, scary, and full of violence,” writes the German digital magazine Bell Tower. Even if for now, AI-generated images are still often recognizable as fakes, it won't stay that way for much longer.
As AI gets better at imitating reality, extremists will be able to tap into a growing set of digital tools to manipulate reality. Der Standard quotes German journalist Sascha Lobo warning against the risk of election manipulation, which goes beyond fake images. Lobo says other terrain being explored is AI-powered audio, which could generate a close imitation of someone's voice to say just about anything.
Germany knows well that, as much as images, words have a long history of inciting the masses.