A snapshot of this is provided by a survey released last week, based on 1,294 professional fact-checks in at least ten languages, conducted by Lupa Agency, a Brazilian outlet specializing in this field. The report is titled The Impact of AI on Global Fact-Checking.

Notícias relacionadas:

• Electoral Court approves restrictions on AI use in 2026 elections.
• AI: Major security concern for businesses in Brazil.
• AI-created fake content in Brazil more than triples from 2024 to 2025.

According to Cristina Tardáguila, innovation and training manager at Lupa, AI is redefining disinformation globally.

“The vast majority of contents analyzed by fact-checkers end up being labeled as false or misleading. AI has rarely been designed to promote truthful content,” she said in an interview with Agência Brasil.

Beyond videos

Another observation by the researcher and founder of Lupa is that misinformation reaches the public in various formats, beyond videos, short audio clips, photos, and text. She also warned about the use of these technologies during election periods worldwide, saying such practices threaten democracies.

“This is an important election year in Brazil and in other countries in the region.” She cites electoral processes in the United States, Peru, Costa Rica, and Colombia. “This scenario will affect both fact-checkers and voters in these countries.”

“They will be inundated with AI-generated content, and there is a high likelihood that much of it is false,” she adds.

According to her, this means that the use of AI to manipulate content is no longer an isolated occurrence but a permanent feature of the digital disinformation landscape. The number of fact-checks identifying this type of falsehood rose from 160 in 2023 to 578 in 2025. By March this year, there had already been 205 such verifications.

Lies in different languages

The study focuses on language rather than geography. In English, 427 cases of AI-generated misinformation and deepfakes, such as face and voice swapping, were identified. In Spanish, there were 198, and in Portuguese, 111.

The researcher argues that the most important priority now is the spread of media literacy. She says fact-checking projects around the world support legislation aimed at helping society recognize potentially false content in posts. Media literacy, she adds, would play a role similar to a vaccine.

“We need the vaccine against misinformation - which is, in fact, quality information - to come first so that people can be prepared and resilient when they encounter AI-driven falsehoods,” Tardáguila emphasizes.

 Cristina Tardáguila from Lupa calls for a public policy that includes media literacy and literacy - the ability to read, write, interpret, and use language effectively - and urges its urgent inclusion in schools - Rawpick/Freepick

Media literacy as a solution

To this end, she calls for public policy that includes media literacy and literacy - the ability to read, write, interpret, and use language effectively - and urges its urgent inclusion in schools.

In addition to the role of government, traditional media companies can also contribute, alongside fact-checking agencies. “It is important to emphasize that fact-checking must adhere to established criteria of transparency and rigor,” Tardáguila notes. The study took into account fact-checks published and indexed by Google’s Fact Check Explorer, a free tool for searching verified information.

“I have no doubt that 2026 will be a year in which we see more and more AI. It is important for Brazilians to know this, to prepare themselves, and to be active and able to identify misinformation,” she says.

The researcher believes that any citizen can fact-check information when they have doubts about its legitimacy. Lupa Agency has created a free course for beginners.