Editor’s note: The AI & Elections Clinic did not conduct this research. This post is part of our effort to highlight emerging work in the field and to give visibility to researchers and practitioners exploring how artificial intelligence intersects with election administration. We share it here to broaden dialogue and support practitioner-informed understanding of this rapidly evolving space.
With input from local election officials, we conducted a national survey of 1,633 U.S. adults (margin of error +/- 2.4% at 95% confidence interval) in October 2025 using AlphaRoc’s Occam platform to understand how the public views the use of AI in election administration, including which applications they find acceptable, how disclosure impacts confidence, and how sources of information impact confidence in election administration.
We find that even if the public accepts AI use under certain conditions, confidence is strongly contingent on source, transparency, and visible human oversight.
Key Findings
Americans approach the use of AI in election administration with tepid openness.
Trust in election communication remains strongly tied to human sources and official channels, and the majority of Americans want clear labeling of AI-generated or AI-assisted content. About three-quarters of Americans say they are very or somewhat confident in information received directly from election officials, and a nearly identical 72% express confidence in information found on official election office websites. Confidence falls to 59% when the information comes from social media posts and to 53% when information is provided by an official election office chatbot.
Attitudes toward AI in election administration vary sharply by age, gender, education, and partisanship. Younger, college-educated, and frequent AI users are more accepting of AI in administrative tasks like translation or answering voter questions, while older adults and infrequent users express greater skepticism.
Across all groups, a majority say it is important for election offices to disclose when AI is used. However, disclosure alone does not necessarily increase trust. Roughly one-quarter of adults say disclosure would increase confidence in their local election officials, one-quarter say it would decrease confidence, and one-third say it would have no effect. Younger respondents and those with a college degree are more likely to report that disclosure would improve confidence, while older and less frequent AI users are more skeptical.
Roughly one-third of adults report encountering false or misleading information about elections through AI tools. These experiences correlate with stronger views, both supportive and skeptical, about the use of AI in election administration.
Overall, the findings suggest that while Americans see potential for AI to improve election administration, maintaining transparency, oversight, and human accountability is essential for sustaining public confidence.
Recommendations
Tailor education and engagement by demographic segment and start with low-risk, high-acceptability tasks. Because attitudes vary based on frequency of usage and demographics, election officials should tailor their rollout and education strategies to particular audiences. Older adults, less frequent AI users, and those wary of automation may need greater explanation, field demos, and trust-building outreach.
Prioritize clear, consistent disclosure and branding. Labels, footers, disclosure pages, and public reports should emphasize human review, process transparency, and how to verify official status (e.g., logo/trust mark).
Pair transparency with accountability. Disclosure is not enough on its own; governance, oversight, audits, and recourse mechanisms must accompany it to convert disclosure into trust.
Address false and misleading information directly. Distinguish AI-assisted official sources from third-party tools, provide easy verification steps, and correct widely circulating AI-fueled errors proactively. Experience with false and misleading information when using AI tools deepens skepticism and shapes demand for protections. Managing confusability between official and non-official AI is critical.
By grounding adoption in public opinion, election officials can ensure that the use of AI reflects both administrative needs and democratic values, reinforcing voter trust in how elections are run and in the public servants who run elections.
