MultiLink_Newsdesk's take
As Americans vote in the midterms, a growing number are asking AI chatbots for election information, and many answers are wrong. An ongoing study by the nonprofit Innovating for the Public Good tested seven major chatbots, including Google's AI tools, Anthropic's Claude, Microsoft Copilot, ChatGPT and Perplexity. In October, 13.9% of answers on voting logistics were outdated or inaccurate, and 5.9% could cause a voter to miss a deadline or vote on the wrong day. Claude and Copilot produced three-quarters of the clear errors. Anthropic said some safeguards may not have been active in the API-based tests; the researchers say web results were comparable. Microsoft says it has improved Copilot. The study also found chatbots often cite prediction markets instead of polls, and campaigns now write content so chatbots repeat it.
Our analysis: Chatbots sound confident even when they are wrong, which makes their errors more dangerous than a bad search result. Companies deserve credit for publishing election policies, but voluntary promises are not enough when a wrong date can cost someone their vote. At the same time, banning chatbots from election topics would not solve the problem, since people would turn to even less reliable sources. The practical lesson is simple: for deadlines and polling places, check official election websites. The deeper question is who is accountable when AI misleads voters.
1 comment