Imagine asking your phone a simple question about the election and getting an answer that is confidently wrong. In Spanish, that is exactly what happened to Laura Zommer, co-founder of Factchequeado, a fact-checking site that fights Spanish-language misinformation. When she asked Meta’s AI chatbot for the date of the U.S. midterm elections, the bot answered November 4 instead of November 3. She asked twice, and twice it gave the same incorrect date. This is not just a tiny glitch in a machine; it is a warning sign. The Institute for Strategic Dialogue later put the same kind of election questions to six chatbots in both English and Spanish. The results were troubling: about 3 in 10 English-language answers were incomplete, outdated, or simply wrong, while 45% of Spanish-language answers were inaccurate. That gap matters because information is power, and voters deserve reliable answers in the language they think and vote in. At an American Community Media national briefing on October 23, Zommer shared these findings with a simple message: artificial intelligence has become a major player in how people learn about elections, and right now, it is amplifying confusion as quickly as it provides answers.
Voters are not waiting for things to improve. They are already turning to AI to research candidates, often with the hope of getting fast, unbiased information. Cameron Hickey, CEO of the National Conference on Citizenship, has been tracking this shift closely. His organization’s polling shows a dramatic jump in the number of eligible voters using chatbots to research candidates as the midterms approach. “We asked that question two weeks ago and the answer nationally of eligible voters was 14%,” Hickey said. “We asked it again this week and it went up to 25%.” But the national number hides a deeper division. Among non-Hispanic eligible voters, reliance on AI hovered around 13%, but among Hispanic voters, that figure shot up to roughly 30%. Why are these numbers so high in some communities and not others? Perhaps because AI feels private, convenient, and free of judgment. Maybe people who have been burned by biased media or who feel left out of mainstream information sources trust a chatbot to be neutral. But when the underlying technology is more likely to fail in Spanish, that trust becomes risky. Voters are making decisions about who will represent them, how tax dollars are spent, and what kind of future they want for their families, and they are making those decisions with the help of tools that are still, in many ways, unreliable. The rapid increase in AI usage should be a wake-up call for everyone who cares about fair and informed elections.
Behind this misinformation problem is something even more disturbing: AI algorithms are designed to hold our attention, and they do that by feeding us messages that enrage us or make us afraid. Hickey pointed to a video circulating online that showed ICE agents intimidating voters at polling sites. The voters in the video were fake, he said, but the images were convincing, in part because real ICE agents have already been seen at polling places, and administration officials have so far refused to rule out deploying agents to polling sites on or before election day. That kind of deployment would violate federal and state law in most cases, but the fear does not care about the law. Another set of videos showed people claiming that touchscreen voting machines were switching their votes. Hickey noted that this exact claim has been floating around for the last three election cycles, and years of investigation have shown it to be extremely unlikely to happen. Yet the video keeps getting shared and reshared because it plays on a deep anxiety: what if my vote does not count? What if the system is rigged against me? These fears are not new, but AI makes them spread faster and look more real. Hickey also described a recent rumor about automated Russian bots supposedly decreasing dramatically after Ukrainian strikes on Russian data centers. The rumor took hold on social media quickly, before Ukrainian President Volodymyr Zelensky even validated it at the Warsaw Security Forum in mid-October, saying there was “less hate, less division and fewer provocations” online because of fewer bots. Part of that claim was true: there were strikes on Russia’s data centers. But, as Hickey explained, “in fact, the same number of bots were there the next day as the day before.” The rumor spread far and wide because people wanted that to be true. Our hopes and fears make us vulnerable to misinformation.
The struggle to find the truth is especially intense in Asian American communities. Jinxia Niu of Chinese for Affirmative Action, which runs the fact-checking site PiYaoBa.org, said that disinformation in the Chinese community has shifted from WeChat to X, formerly Twitter. But WeChat is still extremely important, she explained, because misinformation often starts on X, then gets translated and circulated in WeChat groups and on YouTube. That means falsehoods can cross platforms and languages before fact-checkers even have a chance to respond. Younger people, Niu said, are more adept at spotting misinformation, but immigrant seniors are more vulnerable, especially when they have a high level of education. That might sound surprising, but the pattern is real: the more educated they are, the more time they tend to spend on social media, and the more exposure they have to disinformation. For older immigrants, who may already feel isolated or uncertain in a new country, these false narratives can be especially persuasive. They are not passive consumers; they are searching for guidance in a confusing information environment. Along with PiYaoBa, other reliable sites like VietFactCheck, Tayo, and Savera are working to verify claims for Asian American communities. Their work is essential because misinformation is not one-size-fits-all. It adapts to languages, cultural references, and trusted community spaces. A lie that works in one community may not work in another, and the same is true for facts. That is why community-based fact-checking matters so much.
Perhaps the biggest theme of disinformation this election season is election fraud, and at the center of it is a claim that millions of noncitizens are voting illegally. It is a terrifying story, but it is also almost entirely false. Niu explained that Chinese immigrants on H1B visas, international students, and green card holders have been asking on WeChat groups and Xiaohongshu, a Chinese social media and e-commerce platform, whether their immigration applications could be affected if they mistakenly registered to vote when applying for a driver’s license. Some were worried about software errors or automatic selections on forms. These are real people, with real anxieties, being caught in a net of misinformation that was not designed for them. Hickey pointed out that claims about noncitizen voting are being amplified by what would otherwise be considered trusted messengers: our own government. An internal report by Homeland Security Investigations, which is part of ICE, found that the phenomenon of noncitizen voting is “virtually nonexistent.” Jonathan Blitzer wrote about this in the New Yorker, and extensive research has consistently shown that noncitizen voting has never affected the outcome of an election. Still, the Trump administration remains fixated on illegal voting, with a USCIS quota to investigate 40 names per day. This is not about protecting the integrity of elections; it is about scaring people. The damage goes beyond false accusations: it discourages eligible voters from participating, it fuels distrust in the democratic process, and it makes immigrants feel like they are being watched and judged. For many Chinese immigrants, the fear is personal. They left behind difficult circumstances to build a new life, and now they are being told that a simple mistake at the DMV could put that life at risk. That fear is hard to overcome, no matter how many fact-checks are published.
So what can be done? According to Zommer, fighting disinformation starts with listening. We need to pay attention to the real conversations happening in communities, not just the loudest voices on social media. Fact-checkers and journalists must understand what people are actually asking, worrying about, and sharing before they can correct the record. Hickey agrees, and he advocates for the SIFT method, a simple framework for evaluating online information. The letters stand for Stop, Investigate the source, Find better coverage, and Trace claims back to the original context. Before sharing something that makes you angry or afraid, stop. Do not let your emotions make the decision for you. Investigate who wrote it and why. Look for better, more reliable coverage from established news organizations or fact-checking sites. And trace the claim to its origin to see if it has been altered, taken out of context, or debunked in the past. This method is not perfect, but it gives ordinary people a way to fight back against the flood of manipulation. The panelists at the American Community Media briefing made it clear that the responsibility cannot fall only on everyday voters. Tech companies need to do better, especially in languages other than English. Policymakers need to stop amplifying baseless claims for political gain. Newsrooms, especially ethnic and community media, need resources and support to do their critical work. And each of us, when we see a shocking post or a scary video, can choose to be skeptics rather than shields for lies. Democracy has always required an informed public. In the age of artificial intelligence, that means learning to think like fact-checkers ourselves. We do not need to be experts, but we do need to be curious. We need to ask questions, verify claims, and remember that behind every statistic and every mistake there is a person trying to make a choice about their future. If we can hold onto that humanity, we stand a much better chance of protecting both the truth and the vote.