The dissemination of false content created with artificial intelligence (AI) more than tripled between 2024 and 2025 in Brazil, showing a growth of 308%.
The data is from the first Panorama of Disinformation in Brazilan unprecedented study by Observatório Lupa, which maps trends, targets and the main disinformation tactics. The study was released this Thursday (5).
The study qualitatively and quantitatively analyzed the 617 content verified by the agency in 2025, comparing them to the 839 content in 2024.
The panorama shows that deepfakes and other pieces of disinformation generated with AI went from 39 cases in 2024, a number that represents 4.6% of the total checks carried out by Agência Lupa that year, to 159 in 2025, 25% of checks. This represents an increase of 120 cases.
Deepfakes are technologies that allow faces and voices to be altered in videos, for example, which can generate content with false information.
According to the debut edition of the study, which will be annual, there is a structural change in the disinformation ecosystem.
The research shows that in 2024, AI was mostly used to create digital scams, such as deepfakes of celebrities advertising fraudulent websites, for example. In 2025, technology began to be used strategically as a political weapon: almost 45% of content with AI had an ideological bias, compared to 33% in the previous year.
The Lupa Observatory study identified that More than three-quarters of AI-enabled content circulated in 2025 exploited the image or voice of well-known people, mainly political leaders. The survey points to 36 instances of false content that targeted President Luiz Inácio Lula da Silva; 33, former president Jair Bolsonaro; and 30, minister of the Federal Supreme Court (STF) Alexandre de Moraes.
According to the panorama, the use of WhatsApp to spread disinformation fell from almost 90%, in 2024, to 46%, in 2025. In Observatório Lupa’s analysis, this does not mean that fakes have decreased there, but rather that, now, there is a greater dispersion of platforms.
In addition to Facebook, Instagram, Threads, WhatsApp and X, which were already popular, Kwai and Tiktok, both short video social networks, also became more relevant in the dissemination of fakes.
