Voting rights groups warn of rising AI-generated misinformation in Spanish ahead of the election.

Voting rights groups are raising alarms about the prevalence of misinformation generated by AI models in Spanish, particularly as the U.S. presidential election approaches. A study found that over half of AI-generated responses about elections in Spanish contained inaccuracies, compared to 43% in English. Notable models like Meta's Llama 3 performed poorly, with nearly two-thirds of Spanish responses being incorrect. Advocates stress the importance of verifying election information amid this wave of misleading content aimed at Latino voters.

October 30, 2024
42 Articles

Further Reading