ELECTION OFFICIAL TURNOVER AND MISINFORMATION ARTIFICIAL INTELLIGENCE “DEEPFAKES” On the polling side, too, misinformation threats to a fair 2024 election are worsened by a high turnover of election workers, said William Adler, associate director of the Elections Project at the Bipartisan Policy Center. Administering elections has always been a “relatively thankless, low-paid government job, which has gotten increasingly complex over the past 20 years,” he said. “As we’ve incorporated more technology into polling, they’ve become IT managers in the public spotlight, facing public threats … New responsibilities in how they communicate about their work are a key part of their job as wasn’t the case before 2020 or 2016.” William T. Adler discusses the pressures that are driving many election officials to quit their jobs, which, Adler says, will compromise election safety and integrity. These factors, alongside safety concerns, translate to a high turnover, Adler continued, citing a November 2023 Reed College survey of approximately 1,000 local election officials. In it, “31% of those surveyed said they knew other “local election officials who left their jobs because of personal safety issues and threats. 11% surveyed had considered leaving because of safety concerns, and over a third of them will be eligible for retirement before 2026.” This constitutes a dangerous cycle of misinformation, Adler explained: “Election officials face threats, they may be more inclined to leave their jobs that results in less institutional knowledge on how to run an election, which might result in more mistakes, which may in turn undermine voter confidence, which brings more threats … Delays or errors in processing ballots creates a hunger for information which misinformation peddlers are all too eager to fill.” An increasing amount of this misinformation takes the form of AI-generated “deepfake” images and audio, said Sam Gregory, executive director of Witness.org. This process has gotten particularly easier over the last year, as a wide range of technology has emerged whereby anyone can generate images from a text prompt, or mimic voices from audio samples, “to target and push people out of the public sphere. In conversations I’ve had with people working in electoral processes, this is something they’ve seen and worry about.” Sam Gregory, executive director, Witness.org, shares examples of how the misuse of AI is compromising elections. In electoral contexts, patterns of deceptive image and audio use are already on the rise; he mentioned recent examples of an audio deepfake of Slovakian liberal politician Michal Šime ka and journalist Monika Tódová apparently discussing how to rig the upcoming elections; another audio deepfake targeting UK Labour leader Keir Starmer; and another audio deepfake of Chicago mayoral candidate Paul Vallas. Another form of misinformation involves “cases when someone claims something is a deepfake when it’s actually real,” explained Gregory. Witness. org “receives many cases of deep fakes, and a lot of them are people basically relying on others’ absence of knowledge to deny a piece of audio on the basis that it has been faked.” “Our information environment, by design, discourages engagement and conversation” across opposing political sides, he added. To combat electoral threats like these, we should “start from a baseline of what’s possible, of how misinformation is spread, so we know where to look for and stop it.” Misinformation
13 Publizr Home