← Back to Blog
Research
By ·July 14, 2026·6 min read

Voters Are Asking AI About the 2026 Races. We Scored Every Candidate and 11,000 Government Sites to See Who's Ready.

Something quiet is happening in how people learn about elections. A growing number of voters no longer start at Google or a cable chyron. They ask an AI assistant: who is running for Senate in my state, is this candidate trustworthy, what does this ballot measure actually do. The assistant answers in a paragraph, built from whatever it can find and read on the open web. For the 2026 midterms, that channel is real for the first time, and almost no campaign is measuring it.

So we measured it for them. We took every 2026 federal candidate on file with the FEC - 4,198 of them - and every website in the official U.S. government registry - 16,267 - and ran them through the same AI-visibility scoring we run on millions of businesses. The result is the first dataset of its kind: a report card, per candidate and per government site, on how findable and citable they are to the AI engines voters are starting to trust.

The headline: most of the people and institutions asking for your vote are close to invisible to the AI that voters are asking about them.

What we built

For every site we could reach, we produced a single 0-to-10 AI-visibility score, built from two halves: how cleanly a machine can read the site (structure, clear statements of who and what, the technical signals AI crawlers use) and how much independent authority the rest of the web gives it (news coverage, references, links that carry weight). It is the same methodology behind our 12.8-million-brand research, pointed at politics. No party lens, no endorsements, no opinions - just measurement.

The numbers

The dataset covers 4,198 filed 2026 federal candidates and 16,267 U.S. government websites. Of the government sites, 11,032 return a live, scoreable page today. The rest either sit behind heavy firewalls or are redirect shells - itself a finding, because a site an automated reader cannot cleanly reach is a site an AI cannot cleanly cite.

On the candidate side, the pattern is stark and it is not about fame or money. Down-ballot candidates - state races, challengers, newcomers - are where AI knowledge is thinnest, which means it is exactly where a clean, readable campaign site earns the biggest edge. AI has to lean hardest on what it can actually read when it does not already know who you are. The famous top-of-ticket names are the least affected; the races still up for grabs are the most.

Why this matters for a campaign

When a persuadable voter asks an assistant about a race, three things have to have gone right for a candidate to be named: the AI's crawlers could reach the site, they could read it cleanly, and the wider web gave the model a reason to trust it. Break any one and the candidate is absent from the answer - while an opponent who did the work gets named instead. This is the same dynamic that decides which businesses AI recommends, and it is now arriving in politics.

The encouraging part, and the reason we are publishing this, is that the readable half is fixable fast. It is a set of files and markup with right answers, not a rebuild. A campaign that fixes its site can move from invisible to citable in an afternoon - and almost none of their opponents have started.

The honest limits

This is a snapshot of readiness, not a prediction of who wins. AI visibility is one new factor among many, and we are careful not to overclaim it. Government firewalls block a real share of official sites from any automated reader, so the scoreable set is a floor, not a ceiling. And we score the site's readiness, not the candidate - a low score means the website is hard for AI to read, nothing more. What the data shows cleanly is that the field is wide open and unmeasured, which is precisely the moment to act.

See the board and your own score

We put the marquee races side by side with the polls at engagemii.com/aeo/elections, and we will score any candidate's or organization's website for free, with the exact fixes ranked by impact. Voters are already asking AI about your race. It is worth knowing what it says.

If you want to cite this article, the URL is engagemii.com/blog/we-scored-every-2026-candidate-and-government-site-for-ai-visibility.


Ready to find out if AI can cite your brand?

Get Your Free AEO Score