Utah Valley University student Will Freedman, along with researchers from UVU’s Herbert Institute of Public Policy, Center for National Security Studies, and Vivint SMARTLab, conducted a study on the rise of AI-generated disinformation, known as deepfakes, and their impact on elections. The research found that deepfakes have become increasingly persuasive to the public, with 70% of Americans expressing concerns about AI-generated disinformation affecting battleground states. Notably, a deepfake audio message impersonating President Joe Biden was used in a robocall to discourage voters in New Hampshire.
One key finding of the study was that deepfake videos performed better in terms of credibility, trustworthiness, and persuasiveness compared to real videos. This raises concerns about the ability of viewers to distinguish between authentic and deepfake content. Despite efforts to educate participants about the nature of the study, many were still unable to accurately identify deepfakes, highlighting the challenges posed by this emerging technology.
Researchers emphasized the importance of understanding the impact of deepfakes on viewers and the need for policymakers and campaigns to combat this issue effectively. The study also underscored the potential for malicious actors to use AI-generated content to impersonate individuals, posing significant risks in various scenarios. Moving forward, greater awareness and measures to address the proliferation of deepfakes will be essential to safeguard elections and public discourse.
Source
Photo credit www.deseret.com