Concerns Rise as Musk Shares AI-Generated Video Mimicking Harris

A manipulated video mimicking Vice President Kamala Harris's voice is causing concern about AI's potential to mislead, with Election Day three months away. The video gained traction after Elon Musk shared it on his social media platform X, without clarifying it was a parody. It uses visuals from a real ad by Harris but swaps the voice-over with an AI-generated one.

The video claims Harris is a "diversity hire" and criticises her capabilities. It retains "Harris for President" branding and includes authentic clips of Harris. Mia Ehrenberg, a spokesperson for Harris's campaign, stated, "We believe the American people want the real freedom, opportunity and security Vice President Harris is offering; not the fake, manipulated lies of Elon Musk and Donald Trump."

AI in Politics: A Growing Concern

This incident highlights how lifelike AI-generated media can mislead as the US approaches its presidential election. The accessibility of high-quality AI tools has increased, but significant federal regulation is lacking. This leaves states and social media platforms to create their own rules regarding AI use in politics.

The original creator of the video, YouTuber Mr Reagan, disclosed it as a parody on both YouTube and X. However, Musk's post, viewed over 123 million times, only included the caption "This is amazing" with a laughing emoji. Users familiar with X might know to click through to see the original disclosure, but Musk's caption does not direct them to do so.

Regulatory Gaps and Social Media Policies

Some users suggested labelling Musk's post through X's "community note" feature, but no label had been added by Sunday afternoon. Questions arose about whether Musk's post violated X's policies against sharing synthetic or manipulated media that could deceive or confuse people. The policy does allow exceptions for memes and satire if they do not cause significant confusion.

Musk endorsed former President Donald Trump earlier this month. Neither Mr Reagan nor Musk responded to requests for comment on Sunday. Two experts in AI-generated media confirmed that much of the video's audio was created using AI technology.

University of California, Berkeley digital forensics expert Hany Farid noted the power of generative AI and deepfakes. "The AI-generated voice is very good," he said in an email. "Even though most people won't believe it is VP Harris's voice, the video is that much more powerful when the words are in her voice."

Expert Opinions on AI-Generated Media

Farid suggested that companies providing voice-cloning tools should ensure their services are not misused to harm people or democracy. Rob Weissman, co-president of advocacy group Public Citizen, disagreed with Farid, believing many people would be fooled by the video.

"I don't think that's obviously a joke," Weissman said in an interview. "I'm certain that most people looking at it don't assume it's a joke. The quality isn't great, but it's good enough." He added that because it aligns with existing themes about Harris, many will believe it to be real.

Weissman's organisation has urged Congress, federal agencies, and states to regulate generative AI. He stated that this video exemplifies the kind of issues they have been warning about.

Global Instances of AI Manipulation

Other generative AI deepfakes have aimed to influence voters with misinformation or humour both in the US and abroad. In Slovakia in 2023, fake audio clips impersonated a candidate discussing plans to rig an election just days before voting. In Louisiana in 2022, a satirical ad superimposed a mayoral candidate's face onto an actor portraying him as an underachieving student.

Congress has yet to pass legislation on AI in politics, leaving most regulation to states. More than one-third of states have enacted laws regulating AI use in campaigns and elections, according to the National Conference of State Legislatures.

Beyond X, other social media platforms have policies regarding synthetic and manipulated media. YouTube requires users to disclose if they have used generative AI in videos or face suspension.

The manipulated video raises important questions about how best to handle content that blurs lines between satire and deception as Election Day approaches.