A pro-Ron DeSantis ad used AI to create Donald Trump’s voice. It won’t be the last, experts say.

Never Back Down, a political action committee supporting Florida Gov. Ron DeSantis for the Republican presidential nomination, used former President Donald Trump’s own words against him in a new ad.

Candidates do that all of the time. In this case, however, the ad-makers pushed the boundaries by manipulating audio to read out loud an attack against Iowa Gov. Kim Reynolds in Trump’s voice. 

The message spoken in the ad accurately reflects what Trump wrote on Truth Social, but he did not speak those words himself.

The ad criticized Trump for "attacking" Reynolds, a popular fellow Republican from one of the most important early states in the presidential primary calendar.

The post on Trump’s Truth Social platform said, "I opened up the Governor position for Kim Reynolds, & when she fell behind, I ENDORSED her, did big Rallies, & she won. Now, she wants to remain ‘NEUTRAL.’ I don’t invite her to events!"

(Screenshot)

A viewer wouldn’t know that Trump didn’t say this out loud: Never Back Down took Trump’s words and used artificial intelligence to create audio of a Trump-like voice reading them.

Never Back Down confirmed to PolitiFact that the PAC had "utilized technology to give voice to Donald Trump’s words" from his Truth Social post.

In a statement to PolitiFact, Trump campaign senior adviser Chris LaCivita called "the blatant use of AI to fabricate President Trump’s voice" a "desperate attempt" to deceive the public.

According to Never Back Down, the ad will air in Iowa markets as part of a seven-figure ad buy.

This is not the first time AI helped land a political attack. One DeSantis campaign video showed three AI-generated photos of Trump embracing Dr. Anthony Fauci. Another ad by the Republican National Committee used AI-generated imagery to show dystopian scenes in response to Biden’s reelection campaign.

"Fake audio is nothing new. Fake imagery is nothing new. But AI will let us create much, much more of it," said Colin Delany, a Democratic digital strategist who tracks AI adoption in political consulting for Campaigns & Elections magazine. 

Anjan Mukherjee, a veteran Democratic consultant who is now the founder and principal of Tunlaw Strategies, said he expects to see growing use of AI.

"I do believe there should be some form of oversight on what is and isn't permissible, though it remains to be seen how much of it will happen on its own, and how much might be through Congress," Mukherjee said.

Although the pro-DeSantis ad accurately shares Trump’s words from the Truth Social post, Common Cause disinformation analyst Emma Steiner said she still found the use of AI-generated content in a political campaign "concerning."

"In an information environment where voters are faced with a deluge of false and misleading information, AI content has the potential to muddy the waters even more," she said. "My major concern is that there are, as yet, very few ‘rules of the road’ for the use of AI in political campaigns at a particularly fraught time for democracy — and when social media platforms are backing off of enforcement of their own civic integrity policies."

Evolving regulatory policies and personnel cuts by social media platforms have prompted concerns that more misinformation will proliferate in the 2024 election. For instance, YouTube announced in June that it will no longer take action against videos promoting 2020 election falsehoods.

Digital forensics experts said the audio was likely generated using text-to-speech systems like ElevanLabs that are easy to use. 

The creation process should take less than a few minutes, said Siwei Lyu, a computer science and engineering professor at the University at Buffalo.

Hany Farid, professor at the University of California, Berkeley, said the ad reminded him of a documentary about the late chef, author and TV personality Anthony Bourdain, in which the filmmakers used AI to generate Bourdain’s voice reading an email. Farid found this "ethically dubious but not obviously misleading."

"I would prefer, however, that a campaign (and its supporters) for high office would stay exceedingly far from these ethical lines," Farid said. "Then again, I’m not naive about American politics."

Lyu found the example of the pro-DeSantis ad to be "not hugely misleading" since there is no twisting of the message. But he said there are potentially more problematic uses. 

"When the text is fake and the message harmful, making the real sound of them is more problematic," he said. 

PolitiFact reached out to multiple 2024 presidential campaigns to ask if they had developed a policy on the use of AI, including the way it was used in the pro-DeSantis ad. Only two responded, and they offered divergent positions.

"AI is rapidly changing the way politics is executed. We are using several tools that now have AI components," said Soledad Cedro of the campaign of Francis Suarez, the Miami mayor who is seeking the Republican nomination. 

However, the campaign of Robert F. Kennedy Jr., who is running for the Democratic nomination, said, "The campaign does not use generative AI to put words in Mr. Kennedy's voice, or anyone else's. It is important that when people hear him, they know that it is him."

Some consultants are bullish on the technology, however.

"AI has many applications in our economy and society, and the innovation, discovery, and solutions that come with AI are good," Jamie Burnett, who has worked as a Republican political consultant in New Hampshire. "Authenticity, honesty, decency, and judgment are always going to be the most important ingredients for a candidate and their campaign. Gimmicks, not so much."

There’s little doubt that AI techniques have the political consulting sector buzzing. Delany said he had recently returned from Netroots Nation, a conference for liberal activists, and AI was "a big focus of attention" this year.

The most immediate applications for AI, Delany said, are not necessarily the one showcased in the pro-DeSantis ad. For instance, AI can be used to churn out initial drafts of press releases, social media posts, fundraising emails and ad scripts, saving staffers’ time for final reviews before they are released to the public, he said.

But AI is being scrutinized for audio and visual roles, too, and voters might not be able to detect their use.

"If a tool like Midjourney can create an image for social media or a digital ad faster than I can do it by hand in Photoshop or Canva, that's a productivity boost," Delany said. 

He added, however, that campaigns run risks if they cross the blurry line between what the public will accept and what it won’t, he said.

"Blowback is a real thing, and a campaign that gets associated with false or misleading content may pay the price," Delany said. He said that his biggest concern involves entities who are immune to such public pressure.

"I'm more worried about fake content from outside actors, from individual activists to people trying to make money or make hay off of our political divisions," Delany said. "A Russian troll farm won't face any consequences for messing with American voters."

RELATED: How improperly using AI could deter you from voting, or hurt your health