Authored by Julie Sweet, AAPC Foundation
One month before the 2026 midterms, AI has reached voters faster than it has earned their trust. Voters use it regularly, often without knowing it, but hold it at arm's length on elections and have firm lines for campaigns.
That is the picture from the AAPC Foundation's national survey of 1,568 likely midterm voters, conducted with Inquoro, AMM Political Strategies, and The Tarrance Group from September 14 to 22. It sets a baseline for tracking how voters' use of and trust in AI change as the technology becomes a larger part of how they get political information.
The topline is a mix of rapid adoption, low trust, and clear voter unease about how AI enters elections:
- It's already here. A quarter of voters use AI every day, and a fifth have asked it about elections and politics.
- It often goes unnoticed. Nearly 4 in 10 voters are already getting election answers from AI in their search results, and half of them don't know it.
- Not yet trusted. Only 1 percent of voters consider AI their most trusted source for information about elections and candidates.
- Skepticism runs deepest among the young. Fifty-one percent of voters under 35 say AI hurts people's understanding of elections, and only 19 percent say it helps.
- Optimism splits along party lines. Democrats and Republicans use AI at the same rate but disagree about whether it does more good than harm.
- Voters are setting limits on campaigns. Eighty-four percent reject a campaign cloning its own candidate's voice.
- And they worry most about everyone else. Sixty-six percent are very worried deepfakes will sway other voters; 33 percent are very worried about themselves.
Why this matters now
Social media became a major source of political information long before anyone knew whether people trusted it. Pew Research Center began tracking social media use in 2005 and did not ask about trust in the news people found there until 2016. By then, the platforms were already shaping what voters saw, without a comparable baseline for how trust had developed alongside their use.
In our survey, 44 percent of likely voters got election information from social media in the past month, but only 11 percent call it their most trusted source. We cannot say how that gap opened, because the measurement started too late.
AI is on the same path at an earlier point. The Foundation will repeat this survey as adoption grows, so campaigns, researchers, and policymakers can follow how trust develops as it happens instead of reconstructing it years later.
The questions ahead
Does trust follow use, or does familiarity sharpen skepticism? Will young voters stay skeptical as they use AI more? What happens to trust once AI-generated answers are the norm in search? And will the limits voters set for campaigns hold once synthetic content is ordinary? None will be answered in one cycle; the survey is built to answer them over time.
Review the full findings of the AI Voter Study on the AAPC Foundation website
About the AAPC Foundation
The AAPC Foundation advances research, education, and leadership development to protect political expression and strengthen participation in a changing political and technological environment.