Artificial intelligence | When disinformation threatens the American presidential election

(Washington) Between manipulated images depicting an arrest of Donald Trump and fake videos where Joe Biden announces a general mobilization, the race for the White House in 2024 is threatened by a deluge of disinformation and is seen as the first American election to be hit of the seal of artificial intelligence.


Proponents on both sides are seizing on AI-powered tools, seen by many experts as a double-edged sword.

Campaign teams are using this technology to boost their effectiveness in everything from analyzing election data to writing fundraising emails.

But in an instant, tools make it possible to clone the voice of a political figure, create false videos and fabricate fallacious stories to the point that voters could have difficulty distinguishing truth from falsehood, or destroy confidence in the electoral process.

A video released in June by the team of Republican Florida Governor Ron DeSantis purported to show former President Donald Trump hugging Anthony Fauci, the former White House adviser on COVID-19. However, AFP journalists managed to determine that the video used images generated by AI.

” Far west ”

AI-generated images also showed a false arrest of Donald Trump by police officers in New York. In an equally misleading video, we saw Joe Biden ordering general mobilization to support Ukraine’s war effort against Russia.

“Generative AI threatens to amplify online disinformation campaigns,” writes the NGO Freedom House in a recent report, warning that the technology is already being used to smear reputations in the United States.

Disinformation purveyors employ AI-generated images, audio and text, making the truth easier to distort and harder to uncover.”

Extract from the report by the NGO Freedom House

More than 50% of Americans expect that untruths produced by AI will have consequences on the 2024 presidential election, according to a survey published in September by the media Axios and Morning Consult, an economic research firm.

About a third of Americans said they would have less confidence in results because of AI, according to the same survey.

In a hyper-polarized political environment, some worry that such sentiments will fuel public anger over the electoral process, as when Donald Trump supporters stormed the Capitol on January 6, 2021 on the based on false allegations that the 2020 election was stolen by Joe Biden.

” Revolutionary ”

“Thanks to easy and inexpensive (AI) models to use, we will face a “Wild West” of electoral claims and counter-claims, with limited ability to distinguish fact from falsehood and uncertainties about to their influence on the election,” says Darrell West of the Brookings think tank.

At the same time, rapid advances in AI make it a “revolutionary” resource for understanding voters and trends, points out Vance Reavie of Junction AI.

Previously, teams depended on highly paid consultants to develop strategies and spent hours writing speeches, but artificial intelligence makes it possible to perform these tasks in a flash, he tells AFP .

The authorities are rushing to put in place safeguards, with American states like Minnesota adopting laws to make “deep fakes” aimed at denigrating candidates liable to criminal prosecution.

On Monday, Joe Biden signed a decree to regulate the use of AI.

“Deep fakes use AI-generated audio and video to smear reputations […] spreading false news and committing scams,” said the American president.

The 80-year-old Democrat mentioned seeing a video of himself created from scratch with AI. “I wondered when could I have said that? »


source site-59