AI-Generated Deepfakes Target Brazilian Presidential Candidates, Fueling Election Misinformation

Thumbnail Image

The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.

AI-powered deepfake and deepdub technologies were used to create and circulate fake videos and audio of Brazilian presidential candidates Lula, Moro, and Dilma, depicting them saying things they never did. These AI-generated fabrications have spread misinformation, caused reputational harm, and raised concerns about their impact on the integrity of the electoral process.[AI generated]

Why's our monitor labelling this an incident or hazard?

The event involves AI systems explicitly used to create deepfake videos and audio that impersonate political figures. These deepfakes have been actively disseminated, causing reputational harm and misinformation during election campaigns, which constitutes harm to communities and possibly breaches rights to truthful information. The article references past incidents where such deepfakes were used to damage political candidates, confirming realized harm. Therefore, this qualifies as an AI Incident due to the direct harm caused by AI-generated misinformation in the electoral context.[AI generated]
AI principles
Democracy & human autonomyTransparency & explainabilityAccountabilityRobustness & digital securitySafetyRespect of human rights

Industries
Media, social platforms, and marketingGovernment, security, and defenceDigital security

Affected stakeholders
General public

Harm types
ReputationalPublic interestHuman or fundamental rights

AI system task:
Content generation


Articles about this incident or hazard