US campaigns use AI chatbots for one-to-one voter outreach in 2026

In short: Political campaigns are increasingly using AI chatbots to hold one-to-one conversations with voters at large scale during the 2026 midterms.

What's going on

Campaign outreach is moving beyond mass text messages and toward chat-style conversations that feel personal. Instead of sending the same script to everyone, some campaigns are using AI chatbots, which are computer programs that can “talk” by text, to reply based on what a voter says.

These chatbots are often built to pull from campaign-approved material, like a digital binder of talking points. Some tools can also record what voters care about, so the campaign can track common concerns and adjust its strategy.

Supporters say this can make outreach more responsive. Critics say it can also make persuasion cheaper and more targeted, like having a tireless volunteer who can message thousands of people at once, but who may not always be careful or honest.

What to watch

Several risks are drawing attention. Voters may not realize they are chatting with AI, and a back-and-forth conversation can reveal sensitive political views. Researchers also warn that persuasive chatbots can make mistakes, and stronger persuasion methods have been linked to lower accuracy in some testing, including one older model where over 30% of claims were rated inaccurate.

Evidence that these chats change votes is mixed. One controlled experiment found short chatbot talks could shift candidate preferences, but broader reviews say proof that personalized AI messages are more persuasive than other methods is still limited. Regulators are also catching up, since many rules were written for ads and robocalls, not adaptive conversations that change with each reply.

Source: NYTimes