Germany's Marshall Fund has launched a tool to track AI-generated deepfakes targeting elections circulating in the US and around the world.
tool It comes in a historic election year. More than half of the planet has held or will hold elections in 2024, which will collide with a proliferation of AI-generated audio, video, and images that provide false narratives and information about candidates. This is often the case.
Already in the U.S., from a fake audio purporting to be President Biden urging voters in New Hampshire to sit out the primary, to former President Trump posting a fake image of Taylor Swift falsely suggesting she endorsed him. Until now, we have confirmed that there is a significant amount of AI-generated content targeting elections.
The project, called Spitting Images, only charts deepfakes that have received significant attention or been debunked by journalists.
Lindsey Gorman, director of the project, said the tool will not only provide fact-checking to voters, but also help policymakers consider how to regulate the use of artificial intelligence in elections. I hope that you can discover it.
“We wanted to understand what the current situation is.” [AI] It's actually playing out in the real world in this historic election year. And for policymakers considering potential legislation and guardrails around artificial intelligence, especially political AI, should they have transparency requirements when it comes to politicians and elections? Where should they focus their efforts? Is it? ” Gorman said.
The tracker took data collected last year and charted 133 deepfakes released in over 30 countries.
Gorman said several trends are clearly emerging, including a reliance on audio deepfakes, which account for almost 70% of tracked cases.
“The reality is that the current state of technology is not as convincing when it comes to images and video, but it is even less so when it comes to audio. “It’s difficult,” she said.
AI-generated audio has already played a role in Slovakia's major elections, with fake audio showing one of the candidates, Michal Šimetzka, discussing election fraud and the beer he would drink if he won. They were seen discussing plans to raise prices.
The fake audio appeared during the country's 48-hour campaign suspension, making it difficult to debunk.
Mr Scimeca ultimately lost the election, but it is difficult to know how much of an impact the fake audio had.
However, in some cases, AI reliability may not be an issue.
Elon Musk, owner of social media platform X, shared a photo of Vice President Harris wearing a communist uniform. The photo is obviously fake, but Mr. Musk shared In response to a post by Harris, who called President Trump a dictator and wrote, “Can you believe she's wearing that outfit?”
Another post by Mr. Musk quoted an AI-generated audio of Harris, purported to be a campaign ad, in which Mr. Biden revealed his “bluntness” during the debate and declared himself the “ultimate diversity recruiter.” expressed his gratitude.
Gorman said there should be room for satire and free speech, but even clearly bogus efforts have a role in discourse.
“Even though they know, of course, that Kamala Harris is not going to stand up and say she's an embracer of DEI, they paint that idea in the minds of voters,” she says.
“Repeating the same message over and over again can wear people out. It's a great way for ads like this to keep getting the message across, even when you know it's not real, and to kind of plant seeds of doubt. I think it's one.
There have also been accusations by President Trump that the actual images were manipulated by AI, suggesting that the crowd size at the early Harris rallies was not as large as it appeared. Journalists and other participants pushed back against the false claims by sharing photos of themselves showing large crowds.
Trump's own sharing of an AI-generated photo of Swift also appears to have backfired, citing misinformation about her voting plans when she endorsed Harris last month.
U.S. intelligence agencies are already warning about the role of AI in elections, particularly its use and distribution by foreign adversaries.
Officials with the Office of the Director of National Intelligence (ODNI) said last month that Russia is the top creator of such content, much of it designed to promote divisive issues or create false narratives about U.S. politicians. He said it was aimed at
Iran, on the other hand, primarily uses the tool to create fake news websites, including translating content into Spanish to spread disinformation.
The official noted that not all of the false content shared in recent weeks was created using AI, and that the fake video created using a woman who claimed to have been hit and run by Harris was “staged.” It was pointed out that it was a video of
