Deepfakes in the 2026 elections: challenges, identification, and legal rules

A video that appears to show a candidate making a statement could be entirely fabricated using artificial intelligence—a technology capable of imitating voice, appearance, and movements with remarkable realism.

This warning has gained traction in Brazil. The first Panorama of Disinformation in Brazil, produced by Observatório Lupa, found that false content generated by AI rose from 39 cases in 2024 to 159 in 2025—a 308% increase. The study analyzed materials verified by Agência Lupa over the two years.

In elections, the problem can involve videos, audio, images, and even simulated news reports. In 2024, the Digital Forensic Research Lab (DFRLab) identified 78 confirmed or alleged cases of deepfakes related to candidates in Brazilian municipal elections.

What is a deepfake?

A deepfake is synthetic content created or manipulated using artificial intelligence or equivalent technology to reproduce or alter a person’s image, voice, or expression.

In September 2026, the Superior Electoral Court (TSE) ruled that, for that year’s elections, the definition includes content with a degree of realism or verisimilitude capable of creating, reproducing, or altering the image, voice, or expression of living, deceased, or fictional individuals.

How to identify a deepfake during elections

There is no single indicator that confirms whether content was produced by AI. Verification must consider the material, its source, and the available information about the alleged event.

Several procedures can help:

• Check the source: Find out who originally published the video, audio, or image and verify if the account is trustworthy;
• Compare information: Search for the same statement in news outlets or on the official channels of the person mentioned;
• Analyze audio and visuals: Unusual lip movements, sudden voice changes, unnatural expressions, and visual inconsistencies may indicate manipulation;
• Verify the context: Check the date, location, and circumstances presented in the publication;
• Be cautious of urgent content: Messages encouraging immediate sharing require additional verification.

In the case of audio, the task can be even more challenging. A survey by the German Marshall Fund found that 69% of electoral deepfakes monitored internationally contained AI-generated audio.

Rules for the 2026 elections

The TSE mandates that electoral propaganda using synthetic multimedia content generated or altered by AI must explicitly and clearly disclose this. The rule applies to texts, audio, videos, and images, depending on the format used.

The regulation also imposes specific restrictions on deepfakes used to favor or harm candidacies. Additionally, new synthetic content featuring the image, voice, or expression of candidates or public figures cannot be published, republished, or promoted between 72 hours before and 24 hours after the election, even if labeled.

Why professionals also need to stay alert

For those working with information, recognizing manipulated content is not just an electoral issue. Deepfakes can circulate in professional groups, social networks, and messaging apps, often linked to companies, executives, experts, or institutions.

Before sharing potentially relevant information, verifying the source, context, and seeking independent confirmation reduces the risk of disseminating fabricated content as if it were true.

The advancement of content generation tools makes digital literacy increasingly important for professionals who handle information daily. Understanding how artificial intelligence creates and modifies content also helps recognize its limitations and possibilities.

The goal is not to memorize deepfake characteristics but to develop the habit of verifying information before sharing it. AI tools continue to evolve, and visual cues that currently help identify manipulations may become less apparent over time.

Meanwhile, electoral bodies and digital platforms are adopting their own mechanisms to identify and limit manipulated content. In Brazil, the TSE maintains specific rules for synthetic content and reporting systems related to electoral disinformation.

In the 2026 elections, the combination of more accessible artificial intelligence tools, rapid social media circulation, and specific rules makes verifying the origin of content an essential step before considering information to be true.