Majority of AI Political Influencers in Brazil Fail to Disclose Tech

A study released Wednesday found that 61% ^[2] of AI-generated political influencer avatars in Brazil do not disclose their artificial nature.

This lack of transparency complicates the ability of voters to distinguish between human advocates and synthetic personas during political cycles. The failure to label these accounts violates specific regulations established to protect the integrity of the democratic process.

Researchers identified 18 ^[1] political avatars, referred to as "Donas Marias," active across various social-media platforms. These synthetic influencers are designed to sway political opinion by mimicking human behavior and engagement. Despite the prevalence of these accounts, the majority operate without the mandatory AI-disclosure labels required by Brazil's electoral court, known as the TSE ^[2].

The TSE mandates that any content created or manipulated by artificial intelligence must be clearly labeled to prevent the deception of the electorate. By omitting these labels, the creators of the "Donas Marias" avoid transparency rules intended to curb the spread of synthetic misinformation, a growing concern for regulators worldwide.

The study highlights a gap between the legal requirements for digital campaigning and the actual practices of those deploying AI tools. While the electoral court has set strict guidelines for the use of generative technology, the identified avatars continue to operate in a gray area of enforcement on social media platforms.

These findings suggest that synthetic personas can be deployed at scale to influence public discourse without the friction of legal disclosure. The use of AI to simulate grassroots support or political consensus remains a primary challenge for the TSE as it monitors digital activity on social media.