As the November 3 midterm elections approach, the intersection of technology and democracy has become increasingly fraught, with artificial intelligence, online misinformation, and digital harassment reshaping how voters access information and participate in public life. The stakes are high: elections depend on a shared factual basis and the willingness of citizens to engage, yet both are under strain from sophisticated disinformation campaigns and toxic online environments. Researchers at the University of Colorado Boulder are examining these dynamics with urgency, offering evidence-based insights into how voters navigate a fragmented digital landscape. Jason Thatcher, a professor of information systems at CU Boulder’s Leeds School of Business, studies the effects of artificial intelligence, misinformation, and digital harassment on political participation. His work addresses pressing questions: How do voters respond to AI fact-checking? Why does misinformation persist even after it has been corrected? And how can the fear of political doxing—the public release of private information to intimidate—suppress civic engagement? Alongside him, Susan Abbott, an assistant teaching professor of media studies, examines how AI and social media misinformation affect civic participation, the role of media literacy in helping voters evaluate election information, and approaches to bridging political polarization through citizen assemblies and deliberative polling. Together, their research underscores a sobering reality: the tools that connect Americans to democracy also threaten to disconnect them from it.
Thatcher’s research into AI fact-checking reveals a troubling partisan divide. In a study highlighted by CU Boulder Today, he and his colleagues found that AI fact-checking works, but mostly for progressives. When voters encounter corrections generated by artificial intelligence systems, their response often depends on their existing political identity. Progressives, who tend to trust scientific institutions and algorithmic tools more readily, are more likely to accept the AI’s correction and adjust their beliefs. Conservatives, by contrast, may dismiss the same AI-generated fact-check as biased or untrustworthy, particularly if it contradicts their worldview. This asymmetry has profound implications for election integrity. If automated fact-checking only persuades one side of the political spectrum, it cannot serve as a neutral arbiter of truth in a polarized democracy. Worse, it may inadvertently deepen divisions by giving one group a sense of verification while the other feels attacked. Thatcher also investigates why misinformation persists despite corrections. Psychological research has long shown that once a belief is formed, it is difficult to dislodge, especially when that belief is tied to identity and social belonging. Corrections often fail because they repeat the false claim, reinforcing its familiarity, or because people engage in motivated reasoning, accepting only evidence that confirms what they already want to believe. Social media amplifies this by creating echo chambers where misinformation is repeated and validated. Thatcher’s work suggests that simply debunking falsehoods is insufficient; effective responses must account for the emotional and social dimensions of belief, building trust with communities before attempting to correct them. In a midterm election season rife with fraudulent claims about voting processes and results, these findings are more than academic—they are a warning that the infrastructure of truth is failing precisely when it is needed most.
Beyond misinformation, Thatcher’s research explores how digital harassment, particularly political doxing, can chill civic participation. Doxing involves publishing someone’s personal information—such as their home address, phone number, or workplace—without consent, often to intimidate, silence, or expose them to harassment. In an era of heightened political animosity, doxing has become a weapon of choice against activists, journalists, academics, and ordinary citizens who voice unpopular opinions. A study referenced in CU Boulder Today, titled “Fear of political doxing may push some Americans out of public life,” reveals that the mere threat of doxing is enough to alter behavior. People who fear being doxed may self-censor, withdraw from online debates, decline to attend political events, or even avoid expressing their views in public forums. This chilling effect is particularly pronounced among women, racial minorities, and those with less social or economic power, who face higher risks of severe harassment. The implications for democracy are stark: if citizens are afraid to speak, debate, or organize, the ideal of a participatory political culture erodes. Doxing does not just harm individual targets; it degrades the public sphere by narrowing the range of voices and perspectives available to voters. Thatcher’s work connects this digital harassment to broader patterns of political polarization and democratic backsliding, arguing that the safety of citizens must be a prerequisite for their engagement. He advocates for stronger legal protections, platform accountability, and community norms that discourage doxing, while also urging voters to consider how their own online behavior might contribute to a hostile environment. In the run-up to the midterms, his research serves as a reminder that the health of democracy depends not only on the integrity of ballots but also on the freedom of citizens to participate without fear.
Susan Abbott approaches the same set of problems from the perspective of media studies, focusing on how AI and social media misinformation shape civic participation. She notes that algorithms are not neutral conduits of information; they are designed to maximize engagement, and engagement is often driven by outrage, novelty, and fear. Misinformation that provokes emotional reactions spreads faster and wider than accurate, measured content. During elections, this dynamic can distort voters’ sense of reality, making them believe that fraud is rampant, that certain candidates are pathological, or that their own vote does not matter. Abbott emphasizes that misinformation does not simply confuse people—it actively suppresses participation by fostering cynicism and hopelessness. If voters believe the system is rigged, they are less likely to turn out, volunteer, or engage in constructive dialogue. However, she also points to a remedy: media literacy. Teaching voters how to evaluate sources, check claims, and recognize manipulative techniques can inoculate them against misinformation. Media literacy is not about telling people what to think but about equipping them with the skills to think critically. Abbott advocates for integrating media literacy into schools, libraries, and community programs, and for making it a routine part of civic life. She also stresses the importance of human mediation: while AI can flag suspicious content, trusted human communicators—teachers, religious leaders, local journalists—are often more effective at reaching people across political divides. In an election season where voters are bombarded with information, Abbott’s research offers a practical path forward: instead of relying solely on technological fixes, communities must build the cognitive and social resilience necessary to resist manipulation.
Abbott also explores structural approaches to addressing political polarization, looking beyond individual interventions to the design of democratic institutions. One promising approach is the citizen assembly, a group of randomly selected citizens who deliberate on a specific issue and make recommendations. Unlike traditional town halls, which often attract the most vocal and extreme voices, citizen assemblies are designed to be representative of the broader population. Participants receive balanced information, hear from experts, and engage in facilitated discussions with fellow citizens who may hold different views. This process can reduce polarization by fostering empathy and mutual understanding, while producing well-reasoned policy recommendations that enjoy broader legitimacy. Deliberative polling, a related method developed by political scientist James Fishkin, combines random sampling with in-depth deliberation. Citizens are polled on an issue, then brought together for a moderated discussion, and polled again; the resulting shift in opinion reveals what the public would think if it were well-informed. Abbott argues that these deliberative methods are particularly relevant in an age of misinformation, because they create a protected space where facts can be examined and ideas can be tested without the distorting influence of social media algorithms. She points to successful examples around the world, including citizen assemblies on climate change in the United Kingdom and deliberative polls on electoral reform in Mongolia, as proof that ordinary people are capable of making thoughtful collective decisions when given the right conditions. For the United States, she suggests that incorporating citizen assemblies into the political process could help bridge the partisan divide by giving Americans a shared experience of democratic deliberation. As midterm elections approach, her work offers a hopeful counterpoint to the gloom of misinformation, suggesting that democracy can be strengthened not just by fighting falsehoods but by redesigning how citizens engage with one another.
In a time of deep political divisions, the research from CU Boulder serves as both a warning and a guide. The dangers of AI-driven misinformation, the persistence of false beliefs, and the silencing effect of doxing are real threats to democratic participation. Yet the same scholars who identify these threats also point toward solutions: more effective fact-checking that accounts for partisan psychology, stronger protections against digital harassment, comprehensive media literacy education, and innovative deliberative processes that reconnect citizens across divides. As voters prepare to cast their ballots on November 3, the choices they make will be shaped by the information they consume—and by whether they feel safe enough to participate at all. For journalists covering this election, for educators trying to prepare students, and for citizens trying to make sense of a chaotic information environment, the insights from Jason Thatcher and Susan Abbott offer essential context. They remind us that democracy is not a machine that runs on its own; it requires citizens who are informed, engaged, and unafraid. And they challenge all of us to think critically about the platforms we use, the information we share, and the ways we treat one another in the public square. For those seeking further information or interviews, the contact is [email protected], and additional resources are available through CU Boulder Today. In the final weeks before the midterms, understanding these dynamics is not merely an academic exercise—it is a civic necessity. The future of democratic participation may depend on whether we can harness the power of technology without being consumed by its dangers.