AI Deepfakes Stir Voter Manipulation and Partisan Tensions in US Elections

Photo : AP

A recent surge in disinformation has prompted urgent calls for technology companies to enhance their regulations on generative AI as the U.S. presidential election approaches.

In Washington, the discussion has been fueled by disturbing instances of AI-driven disinformation, including a deepfake video parodying Vice President Kamala Harris, a manipulated clip of President Joe Biden filled with expletives, and a falsified image depicting former President Donald Trump being arrested. These incidents are raising significant concerns about their potential to manipulate voters and intensify partisan divides as the 2024 presidential race heats up.

This election is anticipated to be America's first heavily influenced by AI technologies, with researchers warning that such tech-enabled fabrications could sway voter preferences, discourage participation, and exacerbate political polarization.

Recent events have renewed demands for tech giants to implement stricter measures to regulate generative AI. Many of these companies have previously reduced their efforts to moderate social media content, leading to increased calls for reform.

Stay informed with the latest news. Follow DXB News Network on WhatsApp Channel  

One notable incident involved Elon Musk sharing a deepfake video on X (formerly Twitter) that depicted Harris making disparaging remarks about President Biden. The video, which did not clearly indicate it was a parody beyond a laughing emoji, sparked confusion among viewers. Musk later clarified the video's satirical intent, but concerns remain that it could have misled the public into believing Harris was disparaging both herself and Biden.

In another example, a manipulated video spread across X appeared to show Biden using offensive language against his critics after announcing he would not seek reelection. This video was a deepfake that misappropriated PBS’s logo and was fabricated from footage of Biden's speech denouncing political violence.

Additionally, a deepfake image circulated online purportedly showed Trump being arrested in New York following his conviction for falsifying business records related to a hush money payment. Digital forensics experts confirmed the image was a fake.

Lucas Hansen, co-founder of CivAI, highlighted these examples as indicative of how deepfakes will likely be used in future political contexts. Hansen noted that such AI-driven disinformation is expected to produce images and videos designed to provoke anger and heighten partisan tensions.

Demonstrating the potential for AI to influence voter behavior, Hansen showcased how an AI chatbot could create false tweets claiming that polling locations in Allen, Texas, were charging for parking. Similarly, previous AI-enabled robocalls impersonating Biden sought to suppress voter turnout in New Hampshire's primary.

Further, tests on AI tools like Midjourney have shown the capability to generate misleading images of Biden and Trump. Despite Midjourney’s efforts to block prompts related to these figures, users have found ways to bypass these restrictions, generating unauthorized images with minimal adjustments.

Observers caution that widespread disinformation could undermine public trust in the electoral process. A recent poll by Axios and Morning Consult revealed that over 50% of Americans believe AI-driven falsehoods will impact the 2024 election, with about one-third expecting reduced trust in the election results due to AI misinformation.

In response, tech giants have announced plans to label AI-generated content and address the spread of deepfakes. Advocacy groups have urged these companies to take more decisive action, including banning the use of deepfakes in political advertising and using algorithms to highlight accurate election information.

Free Press, a signatory of an April letter to tech CEOs, criticized the lack of substantive commitments from platforms. Nora Benavidez, senior counsel at Free Press, described the current online environment as toxic, with rampant misinformation complicating the electoral process. She emphasized the urgency for tech companies to strengthen and enforce policies against deepfakes and other deceptive practices to safeguard the integrity of the upcoming election.