In a video clip promoted on social media, an image appears of one of the greatest voices in Brazilian music, in a supposed interview, in which he opens up about his intimate relationship, reveals sexual problems, and ultimately suggests a treatment for erectile dysfunction. “I felt like an 18-year-old boy,” posts the video of therapist Roberto Carlos. Similar products are also sold with fake faces of singer Solimos, player Ronaldo, doctor Drausio Varela and others.
In another ad, the funk singer with more than 13.9 million followers points out a quick and easy way to make money: “What's popular now is Subway Money,” says MC Ryan. But the picture is false.
From promises of a game that could make gamblers millionaires to advertisements for miracle cures for baldness, erectile dysfunction and wrinkles, realistic AI simulation cases are beginning to reach Brazilian courts. It shows how social networks fail to recognize them and keep them off the air.
On the one hand, AI “clones” act as bait for social media users. Public figures, on the other hand, see their images and voices used indiscriminately to sell products and services. The resulting material from “replicating” faces is called deepfakes, a phenomenon that is not new, but has gained momentum with the proliferation of generative AI tools capable of generating content.
Algorithms
On the advertising platform of Meta, which owns Instagram, WhatsApp and Facebook, there are at least 130 results for paid posts from the game Subway Money, which promises players instant money, “no bureaucracy and no fees.” At least since December last year, these posts have featured fake photos of well-known people, such as influencer Virginia and actress Mel Maia, as well as MC Ryan. Keeping it in the air increases meta revenue.
Similar videos are also circulating on TikTok.
While game advertisements focus on the faces of characters popular among young people, advertisements for medicines and medical treatments use images of people that suit all generations. In short videos, deepfakes of Drausio Varela, Ana María Braga, Susana Vieira, Paula Oliveira and Ileana sell the secrets of the “miracle pink capsule” to young people. They all have digitally recreated voices.
Articles are usually short, from 30 seconds to one minute. They carry catchy messages: “You Almost Lost Your Marriage,” “Artists’ Secret to Getting Rid of Sagging, Wrinkles, and More!”
Boosted ads
By circumventing network security systems, enhanced deepfake ads become more powerful. This is because monetization platforms use users' personal and browsing data, using algorithms, to segment audiences and increase the likelihood that a person will click on a particular post, explains Fernando Ferreira, a doctor in computational intelligence and researcher at UFRJ's NetLab:
“When we talk about relevant advertising, this helps companies find their target audience. The problem is that scammers also use such tools.
Ferreira also points out that scammers often use temporary accounts and websites to make tracking difficult. The researcher says the fraud also includes selling non-existent courses and materials and publishing fake e-commerce websites.
This phenomenon has gained momentum on less regulated platforms, says João Finamor, professor of digital marketing at ESPM. While Meta allows you to set active ads, X (formerly Twitter) and TikTok do not offer this capability. It also warns against posting this content in groups on messaging applications.
On their social networks, some people warn “clones” about fake content. But due to the platforms' failure to remove the material, there are those who resort to court.
In April, podcaster and YouTuber Igor Rodriguez Coelho, known as Igor 3K, of Flow Podcast, obtained a positive court ruling against Facebook so that a deepfake that allegedly showed him advertising a fraudulent financial service could be removed, on pain of fine. The decision taken by Ana Paula Mezzina Forlán, of the First Court of the Special Civil Court of São Paulo, recognized the failure of the platform.
The lawyers explained that the broadcaster tried to communicate with him through official channels on Facebook, to no avail. He went to court because he feared damage to his image and scams against Flow users.
responsible
Under the Marco Civil da Internet, social networks are protected from liability for disseminating third-party content – including deepfakes, which are fraudulent bait. Article 19 states that service providers can only be held civilly liable (with fines or compensation) if they do not, after a clear and specific court order, remove or block illegal content within the stipulated period.
Since 2021, the Penal Code considers the penalty for embezzlement (the crime of deceiving someone to obtain a financial or material benefit, by fraud or forgery) more serious when applied in a digital environment.
Since December 2023, Drauzio Varella has filed a civil suit against Facebook, demanding registration data that helps identify individuals or groups spreading deepfakes. In the lawsuit, the lawyers assert that the frauds could constitute crimes against public health, because they involve medications.
In April, the doctor also filed a complaint against Meta with the Public Ministry in São Paulo, arguing that the company was also responsible for allowing the advertisements to be broadcast.
“Coffee trailblazer. Social media ninja. Unapologetic web guru. Friendly music fan. Alcohol fanatic.”