How AI models, including Facebook's, are spreading election misinformation in Spanish

In the days leading up to the presidential election, AI-generated misinformation has surged, disproportionately impacting Latino voters who are vital to both parties' campaigns.

According to an analysis by Proof News and Factchequeado, more than half of AI-generated election-related responses in Spanish contained incorrect information, compared to 43% in English.

Voting rights advocates express concern that the rise of AI in political messaging may exacerbate existing disparities for Spanish-speaking voters, vital in the upcoming election.

Meta's AI model Llama 3 performed poorly, generating nearly two-thirds false information in Spanish responses, demonstrating the urgent need for better practices around AI-generated content.

Read at Fast Company