AI-driven disinformation: The Milei deepfake and the rise of fake news
A deepfake targeting Javier Milei has spread across social media, highlighting how generative AI is flooding the internet with sophisticated misinformation.
English · Original discussion in Spanish · Published
AI-driven disinformation: The Milei deepfake and the rise of fake news
A deepfake attributing a quote to Javier Milei regarding "7,291 foolish elderly people" dying in care homes circulated on social media, shared by profiles with real names and professional websites. This is not an isolated incident: AI-generated images are accumulating millions of interactions, and a significant portion of the public cannot distinguish them. The debate over the quality of the digital sphere is no longer theoretical.
The hoax embraced by those who wanted to believe it
The video, spread via TikTok, showed Milei uttering words he never spoke. The piece was debunked by malditobulo (a Spanish fact-checking organization), but that did not stop a Galician film production company from retweeting it "in outrage" before deleting it. The pattern repeats: fake content does not seek to convince, it seeks to confirm. Those who share it already have their conclusion and only need a visual excuse.
The reaction from the target audience was as expected. One sector attributed it to the left; another, to the far right. The discussion devolved into mutual accusations about who manufactures more hoaxes, without anyone providing conclusive data. The relevant point is that the video worked: it generated indignation, retweets, and real political conversation based on fiction.
AI that cannot distinguish a forum from a cult
The most incisive analysis of the matter was not about the hoax itself, but about the technology. It was noted that AI is trained on human biases and ends up reproducing archetypes: the evangelical, the radical feminist, the local barroom blowhard. "I don't think anyone has bothered to feed any AI to simulate a normal person," one message summarizes. The conclusion is uncomfortable: AI does not surpass us; it copies us.
This imitation has economic consequences. If models generate content based on biased data, the result is a digital ecosystem where verification is optional. Hoaxes are not a system failure; they are its natural product. And the cost is paid by the credibility of any information circulating on social networks.
The business of disinformation
The attention economy rewards the false. A manipulated video generates more interactions than a debunking because indignation travels faster than correction. Algorithms do not distinguish between truth and lies; they only measure time spent on a platform. The result is a market where disinformation is profitable and verification is a cost.
Some argue that the problem is not the bots, but the humans who amplify them. The most repeated analysis points out that there are not as many bots as believed: there is a notable number of university-educated people sharing unverified fake content. This distinction matters because if the problem were only bots, improving filters would suffice. If the problem is human, the solution is much more uncomfortable.
Television that already falsified live broadcasts
Comparisons with classic television appeared several times. It was noted that in the 90s, signals did not reach 500 lines and the brain reconstructed what it could not see. The difference is that now, a generative model performs the reconstruction. The result is the same: the viewer sees what they expect to see.
The reference to Fahrenheit 451 is not accidental. In the novel, television falsifies the capture of the fireman, and the audience applauds in relief. The modern version is a video of Milei that never existed and an audience that shares it as if it did. Fiction no longer competes with reality; it replaces it.
The point where analysis hits a dead end
The discussion remains open. Some call for media literacy; others, for regulation; others, for simply disconnecting. No proposal solves the underlying problem: verification requires time and effort, while digital consumption is designed for the opposite. As long as the cost of creating a hoax is zero and the cost of debunking it is high, the asymmetry will persist. And with it, the feeling that the internet is no longer a place to be informed, but a place to be confirmed.
A deepfake attributing a quote to Javier Milei regarding "7,291 foolish elderly people" dying in care homes circulated on social media, shared by profiles with real names and professional websites. This is not an isolated incident: AI-generated images are accumulating millions of interactions, and a significant portion of the public cannot distinguish them. The debate over the quality of the digital sphere is no longer theoretical.
The hoax embraced by those who wanted to believe it
The video, spread via TikTok, showed Milei uttering words he never spoke. The piece was debunked by malditobulo (a Spanish fact-checking organization), but that did not stop a Galician film production company from retweeting it "in outrage" before deleting it. The pattern repeats: fake content does not seek to convince, it seeks to confirm. Those who share it already have their conclusion and only need a visual excuse.
The reaction from the target audience was as expected. One sector attributed it to the left; another, to the far right. The discussion devolved into mutual accusations about who manufactures more hoaxes, without anyone providing conclusive data. The relevant point is that the video worked: it generated indignation, retweets, and real political conversation based on fiction.
AI that cannot distinguish a forum from a cult
The most incisive analysis of the matter was not about the hoax itself, but about the technology. It was noted that AI is trained on human biases and ends up reproducing archetypes: the evangelical, the radical feminist, the local barroom blowhard. "I don't think anyone has bothered to feed any AI to simulate a normal person," one message summarizes. The conclusion is uncomfortable: AI does not surpass us; it copies us.
This imitation has economic consequences. If models generate content based on biased data, the result is a digital ecosystem where verification is optional. Hoaxes are not a system failure; they are its natural product. And the cost is paid by the credibility of any information circulating on social networks.
The business of disinformation
The attention economy rewards the false. A manipulated video generates more interactions than a debunking because indignation travels faster than correction. Algorithms do not distinguish between truth and lies; they only measure time spent on a platform. The result is a market where disinformation is profitable and verification is a cost.
Some argue that the problem is not the bots, but the humans who amplify them. The most repeated analysis points out that there are not as many bots as believed: there is a notable number of university-educated people sharing unverified fake content. This distinction matters because if the problem were only bots, improving filters would suffice. If the problem is human, the solution is much more uncomfortable.
Television that already falsified live broadcasts
Comparisons with classic television appeared several times. It was noted that in the 90s, signals did not reach 500 lines and the brain reconstructed what it could not see. The difference is that now, a generative model performs the reconstruction. The result is the same: the viewer sees what they expect to see.
The reference to Fahrenheit 451 is not accidental. In the novel, television falsifies the capture of the fireman, and the audience applauds in relief. The modern version is a video of Milei that never existed and an audience that shares it as if it did. Fiction no longer competes with reality; it replaces it.
The point where analysis hits a dead end
The discussion remains open. Some call for media literacy; others, for regulation; others, for simply disconnecting. No proposal solves the underlying problem: verification requires time and effort, while digital consumption is designed for the opposite. As long as the cost of creating a hoax is zero and the cost of debunking it is high, the asymmetry will persist. And with it, the feeling that the internet is no longer a place to be informed, but a place to be confirmed.
Also available in: Deutsch
Summary of a discussion on Burbuja.info - Foro de economía, actualidad y política., translated from Spanish and reviewed before publication. Read the full discussion (103 replies).
More summaries
-
Argentina swaps importers' Central Bank debt for bonds
Milei converts $30 billion in Argentine importers' Central Bank debt into dollar bonds, sparking debate over privatization versus public liability.
-
Huerta de Soto breaks down Milei's ten points from Davos in class
Analysis of Milei's Davos speech: from Adam Smith as a lukewarm social democrat to 58% of income seized by the state, with Argentina footing the bill
-
AI wipes entire disk after programmer asks to clear cache
A developer's request to clear system cache resulted in an AI wiping the entire drive, reigniting debates on human oversight and AI accountability.
-
Debate erupts over Trump's ear and Musk's gesture
Online forums discuss unproven theories regarding a Jewish ritual linked to Trump's ear injury and Elon Musk's recent public greeting.
-
Spanish economy shows growth while cost of living rises
Debate questions record employment figures in Spain as rising food and fuel prices impact household budgets and purchasing power.
-
TVE debates whether pet loss should grant paid leave
Spanish public broadcaster TVE explores if the death or adoption of a pet should entitle workers to paid leave amidst legal uncertainty in Spain.