The term "artificial" may be fitting, but it is worth questioning whether "intelligence" accurately describes computer-generated election misinformation. Election officials have witnessed the damage caused by falsehoods and conspiracies created by humans. The introduction of AI brings the potential for even greater challenges, at an unprecedented pace. Wesley Wilcox, supervisor of elections for Marion County, Fla. , is unfamiliar with AI-generated content infiltrating social media discussions about elections in his state. Even if it did occur, he doubts he would be able to identify it, and he does not believe the state has the resources to invest in developing expertise to combat it. Wilcox states, "My impression is that they don't have the capacity to take on another task — I'd walk that straight out to the jurisdictions as well. " The public has already witnessed glimpses of how AI could be utilized in politics, from fabricated images of the former president embracing Anthony Fauci to deepfake videos of the Ukrainian president instructing soldiers to lay down their weapons. Rick Claypool, research director for the president's office at the nonprofit organization Public Citizen, expresses concerns about the unforeseen consequences that may arise. AI tools have been released without considering their potential misuse, states Claypool. The extent to which they may contribute to distrust in elections and election officials remains unknown, as does the effective response required. Claypool emphasizes, "It shouldn't be up to people at the election official level to sort this out. " Marek Posard, a researcher at the RAND Corporation specializing in countering disinformation, does not view AI as a solitary threat. He is most concerned about its ability to accelerate the forces already eroding trust in elections. Posard asserts, "AI is part of the equation, but the bigger issue is a 'lollapalooza' event where multiple attacks happen simultaneously from foreign actors, domestic actors, and candidates. " Some states have passed bills to address AI's impact on elections, and legislators at both the state and federal levels are working on proposals to mitigate its effects. How might AI create new challenges for election officials? The term "generative AI" refers to AI technology that can produce original content. "Large language" generative AI models rely on vast text databases to generate new written material, ranging from poetry to computer code. Spammers already employ AI to generate blog posts and news stories in hopes of attracting advertising revenues. AI can generate code to build websites, including fictitious "local newspapers, " populated with AI-generated content. AI has the potential to amplify the dissemination of online misinformation and facilitate its spread through email and social media campaigns. These activities require effort but not as much as one might imagine, remarks Renee DiResta, technical research manager at the Stanford Internet Observatory. As an example, she mentions a fake website created by a security researcher as a "proof of concept. " The site automatically generated news articles in response to posts from Russian state media, featuring fabricated writers with manufactured identities and generating tweets about their "stories. " DiResta explains, "It wasn't that hard to make it, but just because you can create something doesn't mean that it's going to have an impact or that it will be effective. " Fake sources still need to attract followers, which presents a significant hurdle in the crowded and chaotic realm of digital influencing. Moreover, recipients must find AI-generated content convincing and persuasive. Cable news channels, social media platforms, and web-based providers of election misinformation with substantial and loyal audiences already exist. The potential exacerbation of the quantity by novel, non-human sources remains to be seen. Gideon Cohn-Postar, legislative director for Issue One, highlights another AI possibility that humans have not yet explored.

Since 2020, election officials have been inundated with public information requests, which they are legally obligated to answer. Many of these requests are boilerplate language created by election deniers. Although this increases the number of record requests, the repetitive nature of the requests has allowed for streamlined responses. Cohn-Postar remarks, "AI would give the possibility of creating many slightly different information requests that could be sent to many different election officials. " Overburdening people and resources that are already stretched thin with complex, unpredictable, and time-intensive tasks is a recipe for trouble, he warns, adding, "It's an area that I think deserves more attention. " AI has the ability to fabricate sights and sounds that never actually existed. This capability becomes particularly concerning in an era when Americans spend an average of over seven hours a day in front of screens and only 30 minutes outdoors in the physical world. AI could create robocall messages from trusted figures, lending an appearance of legitimacy to election misinformation of any kind, including messages that directly suppress voting behavior. Cohn-Postar states, "So far, we haven't heard of a specific instance of AI-generated robocalls. " Another concern lies in deepfake videos, where advancing technology and its widespread availability raise significant apprehensions. While one can still distinguish between an AI avatar and a real person, the gap is closing so rapidly that film and television actors are genuinely worried about being replaced. A deepfake video does not necessarily have to depict an event that never occurred or feature a famous individual to be damaging. Merely adding dimension and realism, thereby increasing plausibility, to any form of misinformation can be detrimental. Misleading video content is already being created by humans through editing, misleading narrations, and broadcasting and posting video footage of inflammatory falsehoods uttered by real individuals. However, generative AI tools could lead to a surge in such content, produced more rapidly and with as-yet-unknown consequences. Similar concerns arise as AI-generated photos become increasingly believable. Posard points out that suspicion surrounding election results is not new, citing examples such as the Bush v. Gore case in 2000. In the past, claims of fraud have emerged, gained popularity for a time, and eventually faded away. Posard worries that AI could generate "evidence" that prolongs controversies based on unfounded claims on social media platforms and television networks. He states, "The technology might create content, but the key is how it might tie content into some kind of cohesive narrative that has a longer time span. " Similarly, Claypool does not want to see the volume of misinformation grow to a point where people become even more distrustful of the entire information ecosystem than they already are. He believes this could impede the functioning of democracy. Wilcox notes the exponential growth in the capabilities and capacity of Generative AI since it became a focal point earlier this year. He remarks, "I think we're just on the front side of what those capabilities are. " Even if states manage to pass legislation to penalize those who employ AI to disrupt elections, Wilcox affirms that criminals do not abide by the law. Likewise, rogue nation-states or stateless actors who wish to cause trouble for American democracy will not adhere to regulations. When asked how he would solve this problem, Wilcox admits, "I don't really know. That's a short answer — I don't know what to do. "