# Views on AI in elections Canonical page: https://data.verasight.io/ai/adults-views-on-ai-in-elections/ Published: 2026-04-29 > Featured topic in AI & Tech. Roughly 56% of adults never use chatbots for political information, and 63% are uncomfortable using AI to help decide a 2026 midterm vote. ## Overview: AI in elections U.S. adults use chatbots for some political information tasks, but discomfort is stronger when AI is used to help choose a candidate. Roughly 56% of adults say they never use chatbots to inform themselves about politics. About 63% say they are not comfortable using an AI chatbot to help decide which candidate to vote for in the upcoming 2026 midterm elections. ## U.S. adults are uncomfortable with AI-assisted vote choice About six-in-ten U.S. adults (63%) say they are not comfortable using an AI chatbot to help decide which candidate to vote for in the upcoming 2026 midterm elections. The largest single response is complete discomfort. Nearly four-in-ten adults (38%) say they are completely uncomfortable with using a chatbot this way. Another 25% say they are somewhat uncomfortable, 25% are somewhat comfortable, and 11% are very comfortable. ## Adults report little political chatbot use Roughly 56% of adults say they never use chatbots to inform themselves about politics. Roughly 24% say they use chatbots to fact-check something they saw online or heard from someone. Another 23% use them to learn more about an ongoing current issue. Some 13% use them to process recent events from the past week. About 8% use them to gather information to win an argument with someone else. ## Older adults are less comfortable with AI vote choice Adults ages 65 and older are less likely than adults ages 30 to 49 to say they are comfortable using an AI chatbot to help decide which candidate to vote for, 22% vs. 48%. Roughly 65% of adults ages 65 and older say they never use chatbots to inform themselves about politics, compared with 47% of adults ages 30 to 49. Among partisans, 59% of Republicans, 63% of Democrats, and 67% of independents say they are not comfortable using an AI chatbot to help decide their 2026 midterm vote. ## Concern about AI bots reaches policy and business surveys Eight-in-ten adults (80%) say they are concerned that AI bots, rather than real people, are answering surveys used to inform government policy and business decisions. This includes 38% who are very concerned and 42% who are somewhat concerned. Roughly one-in-five adults (21%) say they are not concerned. ## Adults reject autonomous-speech treatment for harmful political AI Roughly 45% disagree that false and harmful political content generated by AI should be treated as autonomous speech under freedom of speech principles rather than as a tool used by humans. Another 23% agree that harmful political AI should be treated as autonomous speech, while 31% neither agree nor disagree. ## Supporting evidence ### 63% are uncomfortable letting AI help with vote choice. Question: How comfortable would you be using an AI chatbot to help you decide which candidate to vote for in the upcoming 2026 midterm elections? Note: Anchors the topic in comfort with using AI to help choose a candidate. Topline: - Not at all comfortable: 37.9% - Somewhat comfortable: 25.5% - Not very comfortable: 25.3% - Very comfortable: 11.3% ### 56% never use chatbots for political information. Question: When using AI chatbots to inform yourself about politics, what best describes your reason for doing so? Note: Adds political chatbot use cases, including nonuse, fact-checking, current issue learning, recent-events processing, and argument preparation. Topline: - I never use chatbots to inform myself about politics: 55.6% - Fact-checking something I saw online or heard from someone: 24.3% - Learning more about a current issue that is ongoing: 22.6% - Processing or understanding recent events that happened in the past week: 13.4% - Gathering information to win an argument with someone else: 7.9% ### Adults 65 and older are less comfortable with AI vote choice, 22% vs. 48%. Question: How comfortable would you be using an AI chatbot to help you decide which candidate to vote for in the upcoming 2026 midterm elections? Dimension: age_bucket Note: Anchors the topic in comfort with using AI to help choose a candidate. Comparison: | Measure | 30-49 | 65+ | | --- | ---: | ---: | | Comfortable using AI for vote choice | 47.6% | 21.7% | | Never use chatbots for political information | 47.4% | 64.5% | Topline: - Not at all comfortable: 37.9% - Somewhat comfortable: 25.5% - Not very comfortable: 25.3% - Very comfortable: 11.3% ### 80% of adults are concerned about AI bots answering policy and business surveys. Question: How concerned are you that AI bots, rather than real people, are answering surveys used to inform government policy and business decisions? Note: Adds concern about AI bots answering surveys used for policy and business decisions. Topline: - Somewhat concerned: 41.6% - Very concerned: 37.9% - Not very concerned: 16.6% - Not at all concerned: 3.9% ### 45% disagree and 23% agree on treating harmful political AI as autonomous speech. Question: To what extent do you disagree or agree with the following statement: "When an AI system generates false and harmful political content, it should be treated as an autonomous speaker under freedom of speech principles rather than as a tool used by humans." Note: Adds views on treating harmful political AI as autonomous speech rather than as a tool used by humans. Topline: - Neither agree nor disagree: 31.4% - Strongly disagree: 27.0% - Disagree: 12.6% - Somewhat agree: 8.8% - Agree: 8.4% - Strongly agree: 6.0% - Somewhat disagree: 5.8% ## Methodology - Mode: Verasight panel recruited via random address-based sampling, random person-to-person text messaging, and dynamic online targeting - Population: US adults age 18+ - Field dates: 2026-03-06 → 2026-03-16 - Base (unweighted): 1,000 - Margin of error: +/- 3.4% - Module: Module 1: Technology, Finance, & Media - Sponsor: Verasight - Weight variable: weight - Weighting targets: age, race/ethnicity, sex, income, education, region, metropolitan status ## Sources - [How comfortable would you be using an AI chatbot to help you decide which candidate to vote for in the upcoming 2026 midterm elections?](https://reports.verasight.io/r/omnibus-2026-044#q-1-1): reports.verasight.io/r/omnibus-2026-044: Anchors the topic in comfort with using AI to help choose a candidate. - [When using AI chatbots to inform yourself about politics, what best describes your reason for doing so? (Select all that apply)](https://reports.verasight.io/r/omnibus-2026-044#q-1-21): reports.verasight.io/r/omnibus-2026-044: Adds political chatbot use cases, including nonuse, fact-checking, current issue learning, recent-events processing, and argument preparation. - [How concerned are you that AI bots, rather than real people, are answering surveys used to inform government policy and business decisions?](https://reports.verasight.io/r/omnibus-2026-044#q-1-2): reports.verasight.io/r/omnibus-2026-044: Adds concern about AI bots answering surveys used for policy and business decisions. - [To what extent do you disagree or agree with the following statement: "When an AI system generates false and harmful political content, it should be treated as an autonomous speaker under freedom of speech principles rather than as a tool used by humans."](https://reports.verasight.io/r/omnibus-2026-044#q-1-27): reports.verasight.io/r/omnibus-2026-044: Adds views on treating harmful political AI as autonomous speech rather than as a tool used by humans. ## Citation Verasight Client Omnibus Survey #2026-044, fielded March 6-16, 2026, N=1,000 US adults age 18+, +/- 3.4%. Canonical question URL: [https://reports.verasight.io/r/omnibus-2026-044#q-1-1](https://reports.verasight.io/r/omnibus-2026-044#q-1-1)