Political campaigns try to shape what AI chatbots tell voters

In short: Some political campaigns are publishing content designed to steer what AI chatbots say about candidates, not just what voters see on websites.

What's going on

Campaigns are starting to treat AI chatbots like a new kind of information gatekeeper. Instead of only writing ads and posts for people to read, they are also publishing pages meant to be picked up and repeated by chatbots.

Reuters described a proposed strategy for Massachusetts Republican candidate Mike Minogue. It included creating hundreds of “AI-friendly” policy pages, placing content on websites that chatbots often cite, and then checking chatbot answers to see if they matched his platform.

This approach is often called generative engine optimization, or GEO. In simple terms, it is like search engine optimization, but aimed at chatbot answers instead of Google links. The goal is to make a chatbot more likely to summarize a candidate in the campaign’s preferred words.

This matters because some voters now use chatbots to learn about elections. In an October 2026 Pew survey, 8% of U.S. adults said they used chatbots for 2026 election information. Among those users, 76% looked for candidates’ issue positions and 65% looked for biographical information.

Chatbots can also get basic facts wrong. A Washington Post test of six systems found invented races and misidentified candidates in generated voter guides.

What to watch

The big question is whether voters can tell when a chatbot summary is shaped by campaign material versus independent sources. For now, chatbot answers are best treated like a starting point, like a quick reference, not a final say. Voters can double-check claims against a candidate’s own statements and official materials, and confirm voting rules and locations with state or local election offices.

Source: NYTimes