As the US presidential election approaches, a doctored video impersonating Vice President Kamala Harris' voice has raised major concerns about the misuse of artificial intelligence.
The video, which shows Harris saying things she did not actually say, has sparked a debate about the impact of AI on political misinformation and election integrity.
With Election Day just three months away, the release and rapid spread of this video highlights the urgent need for better rules around AI-generated content.
The deepfake controversy and its impact
The controversial video first gained attention when tech billionaire Elon Musk shared it on his social media platform X on Friday night. In his post, which has been viewed more than 123 million times, Musk called the video “amazing” with a smiley face emoji but made no mention that it was a parody, sparking concerns about AI's ability to mislead people.
The doctored video closely mimics footage from a recent campaign ad by Harris, a leading Democratic presidential candidate, with the original narration replaced with an artificial Harris voiceover.
“The fake voice in the video falsely claims that Harris is a 'diversity hire' and doesn't know how to run the country. While the video maintains the 'President Harris' brand and uses real footage of her, its purpose is clearly to mislead viewers.”
Dr. Kate Tepper, a spokeswoman for the Harris campaign, condemned the video in an email. Associated Press“We believe what the American people want is the real freedom, opportunity and security that Vice President Harris is delivering, not the false, manipulated lies of Elon Musk and Donald Trump,” they said.
AI technology and its role in election misinformation
The incident highlights growing concerns about AI-generated media and its potential to spread misinformation. As AI technology becomes more sophisticated and accessible, the ability to create realistic deepfakes and manipulated media is improving dramatically.
The original creator of the video, YouTuber Mr. Reagan, made it clear that it was meant as a parody, but in response to comments suggesting that manipulating voices in such “advertising” is illegal, Musk responded by saying, “I checked with Professor Sagon Deesnuts, a world-renowned authority, and he said parody is legal in the US.”
This lack of clarity raises questions about the responsibility of social media platforms in managing synthetic media.
Some X users suggested the post be labeled with context, but no such labels had been added as of Sunday afternoon. X's policies prohibit the sharing of synthetic or manipulated media that may mislead or confuse people, but allow exceptions for memes and satire as long as they don't cause “significant confusion about the veracity of the media.”
Musk's post supporting former President Donald Trump earlier this month may not have followed those guidelines.
Reports say AP NewsExperts in AI-generated media have weighed in on the video's impact. Hany Farid, a digital forensics expert at the University of California, Berkeley, praised the quality of the AI-generated audio but stressed the potential for misuse.
“The AI-generated audio is so good,” Farid said. “Most people would not believe it's VP Harris, but having the words spoken in her voice makes the video that much more powerful.”
Public Citizen co-executive director Rob Weissman expressed concern that many people will be fooled by the video.
“I think this is definitely not a joke,” Wiseman said. “I don't think most people who see this would think it was a joke. It's not good, but it's good enough. And it's because it relates to the existing themes surrounding her that most people believe it's real.”
Weissman's comments show how urgent it is to enact rules. Public Citizen has been advocating for Congress, federal agencies and state governments to impose regulations on generative AI to prevent such abuses. Some states have enacted their own laws regarding AI in campaigns and elections, but federal legislation is still pending.
The need for regulatory action
The problem of AI-generated misinformation is not unique to the U.S. Examples of deepfakes have appeared around the world, including fake audio clips in Slovakia and satirical ads in Louisiana. These examples point to the broader challenge of managing AI's influence on the political process.
To address these challenges, social media platforms like YouTube have introduced policies requiring users to disclose the use of generative AI in their content.
However, the effectiveness of these measures remains uncertain, and more extensive federal action is needed to address the rapidly evolving landscape of AI technologies and their impact on democracy.
As Election Day approaches in the United States, the manipulation of AI technology for political gain becomes a pressing issue.
