Eight percent of U.S. adults say they are using artificial intelligence chatbots to find information about the 2026 elections, with candidate positions and fact-checking among the most common purposes.
The finding comes from a Pew Research Center survey of 5,273 adults conducted Oct. 12-18, as voters prepare for the November midterms.
Among people using chatbots for election information, 76% said they sought candidates' positions on issues. Another 71% used them to check the accuracy of things they had heard or read elsewhere about candidates or election issues, and 65% sought information about candidates' lives and careers.
Forty percent of election-information chatbot users sought suggestions about how to choose whom to vote for. The same share used the tools for information about registration, polling locations or voting by mail.
Those percentages describe the relatively small group using chatbots for election information, rather than all U.S. adults. Respondents could report more than one purpose.
Election searches remain a limited use
About half of adults use chatbots such as ChatGPT, Copilot and Gemini for some task, Pew said. The survey's detailed results show that 15% of chatbot users were using them to find information about the coming election.
Election-information use also differed by age: 11% of adults younger than 50 reported doing so, compared with 4% of those 50 and older. Pew said the pattern broadly follows age differences in chatbot use overall.
The findings measure what people say they are using the tools for. Seeking a fact check describes a respondent's purpose, rather than a test of whether a chatbot's answer was accurate.
How the survey was conducted
The respondents belonged to Pew's American Trends Panel, a nationally representative panel of randomly selected U.S. adults. SSRS conducted 5,037 interviews online and 236 by live telephone in English and Spanish.
The survey oversampled non-Hispanic Asian adults and adults ages 18-29, then weighted those groups back to their proportions in the population. Weighting also accounted for sampling and nonresponse and aligned the sample with demographic and other population benchmarks.
The full-sample margin of sampling error was plus or minus 1.5 percentage points. Among the 427 respondents who used chatbots for election information, the margin was wider: plus or minus 5.7 percentage points, according to Pew's methodology.
Eighty-eight percent of the panelists sampled for this survey responded. The cumulative response rate, which also accounts for recruitment and panel attrition, was 3%.
File illustration of artificial intelligence.





