Summary
Image of Lula on fire paired with burning forest goes viral on social media amid record fires in São Paulo’s interior. Candidates and politicians use AI to produce images criticizing the federal government and Lula.
The face of Brazil’s president, Luiz Inácio Lula da Silva (PT), mixed with a burning forest. In the week that the interior of São Paulo recorded a historic record of fires, the image went viral among government critics, who blame Lula for the fire, without providing evidence.
The image, however, was not produced conventionally. It was generated by programs using generative artificial intelligence (generative AI), a type of technology that creates content from text commands submitted by users.
The problem is that the use of images of this type by candidates in this year’s elections, without identifying that they were produced using artificial intelligence, is subject to the veto of the Electoral Court.
In 2024, the Supreme Electoral Court (TSE) updated the rules on advertising, providing, for the first time, two specific points to regulate the use of AI for political purposes. They are: “synthetic multimedia content generated through artificial intelligence must be explicitly informed” and “application providers [como redes sociais ou sites] will be held accountable when they do not promote the immediate unavailability of content and accounts that do not comply with AI labeling methods.”
On X, formerly Twitter, the image has been circulating mainly since August 23, the day that saw the record number of fires in the state. This day alone had more fires than the entire last year in São Paulo, most of which started in just 90 minutes.
On Monday, August 26, São Bernardo do Campo (SP) city councilor Glauco Braido, MDB, candidate for re-election, published the image. Despite the resolution on TSE, there is no label in the post indicating the use of AI. The publication only states that it is electoral propaganda material from the coalition “São Bernardo Que Segue em Frente” (PP/PSD/Solidariedade/PL/MDB/PSDB/Cidadania). The name of the candidate and the caption number also appear in the image, with the overlay “O GOVERNO DAS QUEIMADAS!”.
In the caption of the post, Braido, who identifies himself as a member of the right-wing movement Brasil Livre (MBL), asks: “What is your opinion on the role of Lula and Marina Silva in fighting the fires?” Then, ask for votes and reveal your ballot box number. The post has received almost a thousand likes and 20 comments.
Also from the MBL and who published the image (in video) on the same day was the candidate councilor of Americana (SP) Rafael Macris, of Progressistas. As with Braido, there is no label indicating that the image was produced using artificial intelligence, only the mention of the caption of the coalition “Americana no Topo de Novo” (Progressistas/União Brasil/Cidadania/MDB/PSDB).
In the publication, Macris, who presents himself as an agri-food entrepreneur, asks: “What do you see in this video?” In the comments, one user replies, “the face of his partner is on fire.” In another, “the image and likeness of the devil LULA himself!!!.” The post has had 649 likes and 42 comments.
He had already posted the image on another occasion, on March 6, even before the start of the electoral propaganda period. In the publication, which had 143 comments, Macris mentions Braido and writes: “She is on fire with so much love!”
According to Fernando Neisser, a lawyer specializing in electoral law, the lack of a label on the seats indicating the use of artificial intelligence constitutes an electoral irregularity. “It is a behavior that can lead to the imposition of a fine or the revocation of the candidacy, but it does not lead to prison, it is not a crime,” he explains.
The story of the image of Lula mixed with fires begins much earlier, a year ago. It appeared on one of the largest Brazilian political discussion forums on Reddit, a free platform that works like a forum, where you can create groups, write and respond to posts.
Brasil Livre has more than 218,000 members and presents itself as “the freest Brazilian subreddit on reddit.” It frequently posts criticisms of the Workers’ Party, Lula, and the Minister of the Supreme Federal Court (STF) Alexandre de Moraes.
On August 13, 2023, the first post of the group appears. The person responsible is an active user of Brasil Livre, but whose account has now been suspended by Reddit itself. He shared four images produced by Lula’s face. Two of them mixed scenes of forest fires with the president’s face. The other two combined Lula’s face with images of favelas.
The post contained the instructions: “Final compilation to send to Zap’s aunts!” In one of the comments, the user explains how the creation took place: “The process is done using Stable Diffusion + Control Net. There is a lot of material about it on YouTube and Reddit.”
Stable Diffusion is an open source program created by the startup company Stability AI, owned by British/Bangladeshi tech entrepreneur Mohammad Emad Mostaque. The program was launched in August 2022, catapulting the company, which was valued at $1 billion.
The program, however, has been accused of allowing users to customize the code and use it to generate fake, lifelike images of naked people, so-called nude deepfakes. One of the most high-profile recent cases was that of pop singer Taylor Swift, who created fake images using artificial intelligence. Images of armed politicians have also been generated by users, bypassing security blocks in the program’s original code.
The origin of the image with Lula had already been verified by the control vehicle Factchequeado, coming from Latin America, in 2023.
The irregular use of artificial intelligence has already led to a fine for one candidate
As Neisser explains, despite the TSE ban, only other candidates, parties, coalitions and the Public Election Ministry will be able to ask the Electoral Court to assess the use of AI-generated images shared without the appropriate label. The same rule applies to other types of publications prohibited by the Electoral Court, such as disinformation content in general or the use of chatbots to mediate communication with voters.
The deadline is two weeks for the electoral representation to analyze the justifications of the interested party and notify the possible offenders, who can present their defenses. Then the judge decides, being able to order the removal of the links and posts from the platforms, with an appeal in one day for both parties – the one who filed the complaint and the one who is reported.
In May of this year, the mayor of Guarulhos (SP), Gustavo Henric Costa, a Republican candidate for re-election, was fined R$ 5,000 by the electoral court after a publication made using artificial intelligence on Instagram. The court found that the post constituted early electoral propaganda against the PT. The decision of the electoral judge Gilberto Azevedo de Moraes Costa, of the 278th electoral district of Guarulhos, highlighted the electoral irregularity of the content, in which, according to him, “the manipulation of the image is evident given the image defects characteristic of the use of this tool”.
The use of artificial intelligence in elections has been analyzed by Desinformante, a platform that brings together researchers on disinformation. Tatiana Dourado, a researcher at the National Institute of Science and Technology in Digital Democracy (INCT.DD), and Democracia em Xeque, spoke about the report. According to her, these publications can increase disinformation by activating senses such as humor.
“It is by mixing humor with radicalized politics that extremist politicians and narratives expand the audiences they wish to co-opt. From an electoral perspective, this is entirely strategic. What purpose does humor serve in elections if not to mobilize?” he asks.
In the case of the images produced by AI that associate the fires with the federal government, Dourado explains the mobilization generated in public opinion based on the ambiguity suggested by the image. “That is why it is risky for the TSE to promote partnerships with the control bodies only during elections, because at other times the problem becomes naturalized. It is from doubt that disinformation arises and that it undermines the judicial systems or the press itself. The use of the AI mass replication technique is precisely to break this disinformation fight by the institutions,” he says.
Politicians Jump on AI Bandwagon to Criticize Federal Government
It is not only the council candidates who are broadcasting images produced by AI that associate the fires with the federal government. On August 26, the federal deputy José Medeiros (PL-MT) published two images produced using AI on his personal Instagram account. One of them represents the map of Brazil in flames, covered in smoke. The publication has 67 comments, to all of which Medeiros responded with a fire emoji.
In the other image, in which the acronym MST appears engulfed in flames in the middle of a forest, the deputy writes: “Only God is responsible!” The post has 351 comments and follows the same pattern of interaction with followers as the other publication, only changing the emoji in the reply.
The same image was also shared by federal deputy Zé Trovão (PL-SC). As reported by Agência Pública, this week rural deputies used their social networks to spread false news accusing the Landless Rural Workers Movement (MST) of being behind the forest fires.
Federal MP José Medeiros (PL-MT) posted two AI-generated images on his personal Instagram account
On August 28, Senator Damares Alves (Republicans-DF) published a video produced using artificial intelligence. The publication mixes current news headlines from newspapers such as Revista Oeste, Folha de S.Paulo, BNC and BBC with a video from 2021 in which artists such as Gilberto Gil and Daniela Mercury call for the saving of the Amazon. In the montage, the singers appear involved in a fire. In the caption, Damares writes: “Record number of fires. And imagine if it were under the Bolsonaro government.” It received 1,847 comments.
Source: Terra
Rose James is a Gossipify movie and series reviewer known for her in-depth analysis and unique perspective on the latest releases. With a background in film studies, she provides engaging and informative reviews, and keeps readers up to date with industry trends and emerging talents.
