A Manipulated Video Shared By Musk Mimics Harris’ Voice, Raising Concerns About AI In Politics

NEW YORK — A modified video that mimics Vice President Kamala Harris’ voice saying things she did not say is raising concerns about artificial intelligence’s ability to deceive, with Election Day just three months away.

The video gained traction after tech entrepreneur Elon Musk published it on his social media site X on Friday evening, without clearly stating that it was originally released as a spoof.

The video features many of the same images as a genuine ad Harris, the likely Democratic presidential contender, published last week to launch her campaign. But the video replaces the voice-over audio with another voice that convincingly impersonates Harris.

According to the video’s narration, “I, Kamala Harris, am your Democratic candidate for president because Joe Biden finally exposed his senility at the debate.” The video argues that Harris is a “diversity hire” because she is a woman and a person of colour and that she doesn’t know “the first thing about running the country.” The video retains the “Harris for President” branding and includes some genuine past clips of Harris.

Musk | AP News Image

A Manipulated Video Shared By Musk Mimics Harris’ Voice, Raising Concerns About AI In Politics

In an email to The Associated Press, Harris campaign spokesperson Mia Ehrenberg stated, “We believe the American people want the real freedom, opportunity, and security that Vice President Harris is offering; not the fake, manipulated lies of Elon Musk and Donald Trump.”

The widely circulated movie exemplifies how lifelike AI-generated images, films, or audio excerpts have been used to mock and mislead politics as the United States approaches the presidential election. It demonstrates how, despite high-quality AI technologies having grown significantly more available, there has been little federal effort to restrict their use, leaving AI in politics to be mostly governed by states and social media companies.

The film also raises concerns about how to effectively handle content that blurs the borders between what is deemed a suitable use of AI, especially if it comes under satire.

The original individual who posted the video, a YouTuber named Mr Reagan, has stated on both YouTube and X that the modified video is a parody. But Musk’s tweet, which has been watched over 123 million times, merely carries the phrase “This is amazing” and a laughing emoji.

X users who are experienced with the site may know how to navigate from Musk’s post to the original user’s post, where the disclosure is available. Musk’s caption does not ask them to do so.

While some users of X’s “community note” option to offer context to posts recommended labeling Musk’s post, no label had been added as of Sunday afternoon. Some individuals online questioned whether his tweet violated X’s principles, which state that users “may not share synthetic, manipulated, or out-of-context media that may deceive or confuse people and lead to harm.”

The policy excludes memes and satire as long as they do not create “significant confusion about the authenticity of the media.”

Musk backed former President Donald Trump, the Republican contender, earlier this month. Mr Reagan and Musk waited to respond to emailed requests for comment Sunday.

Two experts in AI-generated media analyzed the false ad’s audio and confirmed that much of it was created with AI technology.

Hany Farid, a digital forensics expert from the University of California, Berkeley, believes the movie demonstrates the capabilities of generative AI and deepfakes.

“The AI-generated voice is very good,” he stated via email. “Even though most people won’t believe it is VP Harris’ voice, the video is that much more powerful when the words are in her voice.”

He stated that generative AI businesses that make voice-cloning and other AI tools available to the public should do more to ensure that their services are not used in ways that endanger people or democracy.

Rob Weissman, co-president of Public Citizen’s advocacy group, disagreed with Farid, believing that the film would mislead many people.

“I don’t think that’s a joke,” Weissman explained in an interview. “I’m sure most people looking at it don’t think it’s a joke. The quality could be better, but it’s adequate. Most people will believe it is true because it feeds into the themes floating around her.

Weissman, whose organization has lobbied for Congress, federal agencies, and states to govern generative AI, described the video as “the kind of thing that we’ve been warning about.”

A Manipulated Video Shared By Musk Mimics Harris’ Voice, Raising Concerns About AI In Politics

Other generative AI deepfakes in the United States and worldwide might have attempted to persuade voters using misinformation, humor, or both. In Slovakia in 2023, phony audio samples depicted a candidate discussing intentions to rig the election and boost beer prices days before the vote. In 2022, a political action committee created a parody advertisement in which a Louisiana mayoral candidate’s visage was placed on an actor presenting him as an underachieving high school student.

Congress has yet to approve laws on AI in politics, and federal agencies have taken only limited steps, leaving the majority of existing U.S. regulation to the states. According to the National Conference of State Legislatures, more than one-third of states have enacted legislation governing the use of artificial intelligence in campaigns and elections.

Aside from X, other social media firms have procedures to address synthetic and modified media published on their networks. For example, users of the video platform YouTube must disclose whether they utilised generative artificial intelligence to make films or risk suspension.

SOURCE | AP