Political candidates have a new audience to court, and it does not vote: the artificial intelligence chatbots that millions of Americans are quietly asking about the 2026 races.

A cottage industry known as answer engine optimization is now selling campaigns, dashboards, audits, and tactical fixes to shape what ChatGPT, Google's Gemini, and other large language models say about candidates, and the tools are already changing how campaigns manage their web presence.

Run for Something Action Fund launched CampSight last month, giving progressive candidates a live look at how chatbots describe them and their opponents.

The premise is straightforward.

Modern chatbots no longer rely solely on stale training data; they pull fresh material from the open web to answer questions about current events, including candidates.

That means a well-timed Wikipedia edit, a rewritten policy page, or a Reddit thread can end up quoted back to voters within minutes.

Caucus AI, a firm co-founded by former Kamala Harris campaign analytics chief Meg Schwenzfeier, ran an experiment that found content newly added to Wikipedia appeared in chatbot answers within about 12 minutes.

The stakes are not academic.

Demos, a British think tank, tested five popular AI services during the pre-election window ahead of Scotland's May 2026 parliamentary vote and found that 34.1% of responses about three real constituencies contained factual errors, including invented candidates, fabricated scandals, and wrong voting dates.

Replika performed worst at 56.4%, ChatGPT next at 46.2%, with Gemini and Grok well behind.

In companion polling, Demos said one in five U.K. adults reported using an AI tool for election information ahead of the May 7 vote.

American operatives are watching similar behavior spread.

Dustin Lloyd, a Democratic state House candidate in Missouri, described using CampSight to rewrite policy pages and improve site coding so chatbots would surface his small-business platform.

Amanda Litman, a Run for Something cofounder, said the tool aims to close the gap between how candidates describe themselves and how AI describes them.

Caucus AI is now tracking chatbot output on Senate, House, and gubernatorial races and has found that former Alaska Rep. Mary Peltola, a Democrat, has had her "Fish. Family. Freedom." slogan, central to her Senate challenge to Sen. Dan Sullivan, R-Alaska, picked up consistently across models.

The influence cuts both ways.

Tim Chambers of Dewey Square Group, whose firm published research in February, said conservative, far-right, and low-factuality sites allow AI crawlers more freely than center-left or high-factuality outlets, structurally skewing what models ingest.

Disinformation actors, he warned, will move fastest to exploit the same optimization playbook now being marketed to candidates.