Campaigns brace for misleading answers from voter AI chatbots

In short: More voters are asking AI chatbots about candidates and elections, and experts warn this can spread both accidental mistakes and deliberate, realistic-looking lies.

What's going on

Voters are increasingly using generative AI tools, like chatbots built into search and Q&A features, to learn about candidates’ positions, backgrounds, and even how and where to vote. Researchers and election-law experts say these systems can “hallucinate,” which means they sometimes make up details when they do not have clear information (like a confident student guessing on a test).

That creates a new problem for campaigns. A chatbot is a third-party source that campaigns do not control, yet it can shape what voters think. For lesser-known local candidates, the risk can be higher because there is less reliable information online, so the AI has more gaps to fill.

At the same time, AI can also be used on purpose to trick voters. Recent examples include deepfakes, which are fake but realistic video or audio, and an AI-made robocall in New Hampshire that impersonated President Joe Biden and told people not to vote. Experts warn that fake ads and synthetic media can be produced quickly and cheaply, and can be tailored to specific groups, which makes them harder to spot and respond to.

What to watch

Some US states now require labels on political deepfakes or restrict them close to Election Day, but these laws usually do not cover everyday chatbot answers about candidates. For voters, the safest approach is to treat AI answers as a starting point and double-check voting rules on official election websites and major news sources.

Source: NYTimes