“This isn't just a media problem. This is a civic health problem,” said Cameron Hickey, CEO of the National Conference on Citizenship, warning that misinformation and disinformation are eroding public trust in democratic institutions.
As the November 3 midterm elections approached, researchers and fact-checkers warned that artificial intelligence, misleading political narratives, and immigration-related fears were creating new challenges for voters, particularly within immigrant and non-English-speaking communities.
During an October 23 media briefing hosted by American Community Media, three experts discussed how false election information spreads through social media, private messaging groups, and increasingly, AI-generated content.
The panel featured Hickey; Jinxia Niu of Chinese for Affirmative Action; and Laura Zommer, co-founder and CEO of Factchequeado, a nonprofit organization combating Spanish-language misinformation in the United States.
Speakers emphasized that disinformation is not simply a matter of false information circulating online. It can influence whether eligible citizens feel safe voting, how communities understand election laws, and whether voters trust that their ballots will be counted.
Hickey explained that misinformation has existed for generations, but technology has fundamentally changed how quickly false narratives can spread.
A rumor that once took weeks to circulate can now reach thousands of people almost immediately. Artificial intelligence has also made it easier to manufacture convincing images, videos, and political messages.
Hickey described examples of AI-generated images depicting Immigration and Customs Enforcement (ICE) agents intimidating voters at polling locations.
Although the depicted incidents were fabricated, he said such imagery could heighten fears among immigrant communities, particularly naturalized citizens already concerned about immigration enforcement.
He also warned that misleading claims about noncitizen voting were being amplified through political messaging and government statements.
One concern involved the federal Systematic Alien Verification for Entitlements (SAVE) database, which Hickey said was being used to question voter eligibility despite potential inaccuracies involving newly naturalized citizens.
According to Hickey, outdated citizenship records could result in eligible voters being incorrectly identified as noncitizens.
He emphasized that election misinformation can discourage participation even when those spreading it do not intend to mislead.
“People are scared to vote, people are scared to participate, people make decisions that are different based on that mis- and disinformation,” Hickey said.
Jinxia Niu, who leads the Chinese Digital Engagement Program at Chinese for Affirmative Action, described how election disinformation is circulating within Chinese-speaking communities.
Her organization has tracked more than 600 pieces of disinformation and identified a significant shift in where misleading narratives originate.
While WeChat previously served as a major platform for Chinese-language election misinformation, Niu said X, formerly Twitter, has become an increasingly important source.
False narratives frequently begin in English, circulate on X, and are subsequently translated into Chinese before spreading through WeChat groups and YouTube.
Niu estimated that approximately 60% of the Chinese-language disinformation her organization monitors is translated directly from English-language right-wing narratives.
She identified election fraud allegations, noncitizen voting claims, and anti-immigrant messaging as recurring themes.
Niu explained that cultural differences can make certain false narratives particularly persuasive. For example, voters familiar with countries where identification is universally required may misunderstand the decentralized voting systems used in the United States.
Claims that American elections allow anyone to vote without verification can therefore exploit unfamiliarity with state-specific election procedures.
Niu also warned that immigration-related disinformation has generated fear among naturalized citizens who are legally eligible to vote.
“This fear has, we can already sense it has already been keeping eligible citizens home, too,” she said.
Her organization has also documented similar anti-immigrant narratives affecting Indian, Filipino, and Vietnamese communities.
Laura Zommer of Factchequeado described a sharp increase in AI-generated political content during the 2026 election cycle.
She cited research from the Wesleyan Media Project identifying at least 169 political advertisements using artificial intelligence by early September, representing approximately $80 million in advertising spending.
According to the research she presented, 69% of those advertisements did not disclose that AI had been used.
Zommer explained that while some states have enacted laws governing AI-generated political advertisements, the regulations vary significantly, and there is no comprehensive federal requirement for political campaigns to disclose AI use.
She said the increasing sophistication of synthetic videos makes it more difficult for voters to distinguish authentic political messages from manipulated content.
The problem is particularly concerning for communities that receive much of their political information through social media rather than traditional news outlets.
Beyond political advertising, the speakers raised concerns about the growing use of AI chatbots as sources of election information.
Zommer cited research by the Institute for Strategic Dialogue examining responses from six major AI chatbots to questions about U.S. elections.
According to the findings she presented, approximately 30% of English-language responses were incorrect, incomplete, or outdated.
In Spanish, that figure increased to 45%.
“The gap in the language is also present in the AI models and the chatbots,” Zommer said.
Niu raised similar concerns about Chinese-language chatbot responses, explaining that searches conducted in Chinese can surface different information sources than equivalent English-language searches.
She warned that AI systems may direct users toward unreliable or highly partisan sources, especially when trustworthy reporting in their preferred language is limited.
Hickey also described preliminary polling conducted by his organization suggesting that the proportion of eligible voters using AI chatbots to learn about political candidates had risen from 14% to 25% over a two-week period.
He said the findings highlighted the growing importance of accurate, accessible election reporting that AI systems can reference.
Despite the technological challenges, the speakers agreed that trusted community relationships remain among the most effective ways to counter election disinformation.
Niu encouraged journalists to work with community leaders, local organizations, and individuals who already have established relationships with their audiences.
She explained that fear-based misinformation is particularly difficult to counter because people are often more likely to believe messages shared by friends, relatives, or familiar community figures.
Zommer described Factchequeado's network of 151 media and grassroots organizations, which work together to produce fact-checks, explainers, and media literacy resources for Spanish-speaking audiences.
She emphasized that journalists must also listen directly to community members to understand which rumors are circulating through private conversations and messaging platforms.
Both Zommer and Niu stressed that vulnerability to misinformation is not limited to people with less education.
Niu noted that even highly educated immigrants can struggle to identify misleading information without specific training in fact-checking and media literacy.
“Fact checking and media literacy is a very, very specialized skill,” she said.
Hickey encouraged journalists and voters to use the SIFT method developed by information literacy researcher Mike Caulfield: Stop, Investigate the source, Find better coverage, and Trace claims to their original context.
The approach encourages individuals to pause before sharing emotionally provocative content and verify information through reliable sources.
Speakers also urged community media organizations to publish accurate election information in multiple languages, particularly on subjects such as voter eligibility, ballot access, immigration enforcement, and election security.
Zommer emphasized that correcting misinformation does not always change a person's beliefs, but it can change their behavior, including whether they choose to share misleading content.
As AI-generated political messaging becomes increasingly widespread, the panelists argued that reliable local journalism, multilingual fact-checking, and trusted community networks will remain essential to protecting informed participation in elections.
The challenge, they said, extends beyond any single election: ensuring that voters can distinguish reliable information from manipulation in an increasingly complex digital environment.
