AI chatbots are giving voters wrong answers ahead of the November 3 US midterm elections, with startup Forum AI finding that almost 40% of chatbot responses to election-related questions contained at least one factual error, AFP reported Sunday. The vote is the first federal election in which millions of Americans are actively relying on AI tools for political information.
Forum AI, a startup that evaluates AI models, examined eight models and found that questions about voting logistics were answered incorrectly more often than questions about candidates. The startup said chatbots put the wrong words in the mouths of candidates, officials and observers hundreds of times, and frequently cited low-quality sources such as Reddit threads.
Forum AI's examples were specific. When its researchers asked Anthropic's Claude why voters should elect Representative Don Davis of North Carolina, the model listed "former school superintendent" among his credentials, confusing him with a different person. Google's Gemini 3.6 Flash, asked about voting in Pennsylvania, gave November 3 as the deadline to request a mail-in ballot. The actual deadline was one week earlier, and Forum AI noted that a voter who followed the answer would have requested a ballot too late to use it.
Key facts
- 01Forum AI found almost 40% of chatbot answers to election questions contained at least one factual error across eight models.
- 02Google's Gemini 3.6 Flash gave November 3 as Pennsylvania's mail-in ballot request deadline; the real deadline was a week earlier.
- 03NewsGuard found Google AI Overview, Gemini and Microsoft Copilot repeated a fabricated claim about Texas Senate candidate James Talarico.
- 04Two in five Americans used AI tools for election information, per a National Conference on Citizenship survey.
- 05A Brennan Center study found ChatGPT, Gemini, Grok, Claude, Perplexity and DeepSeek disputed every election conspiracy theory tested.
Anthropic did not respond to AFP's request for comment. Google said it is focused on helping voters find reliable information across its platforms and that it applies strict quality and policy protections to election topics across its AI products.
The chatbots also stumbled on a foreign influence campaign. In late October, AFP fact-checkers debunked a likely AI-altered clip purporting to show a former student of Texas Senate candidate James Talarico accusing him of sexual misconduct. The unfounded claim bore the hallmarks of a Russian disinformation campaign. Disinformation watchdog NewsGuard found that Google's AI Overview repeated the allegation when prompted, and that Gemini and Microsoft's Copilot repeated it as well.
Both companies said their systems corrected course. Google said the issue was addressed because its systems quickly updated to surface the latest reporting on the topic. Microsoft said it was a rapidly developing situation with limited reliable information available at the time of the initial query, and that Copilot later reflected the updated context and identified the claim as fabricated. A later AFP test, run after the claim was widely debunked, found the tools citing professional fact-checkers.
In a separate study of 10 AI tools, NewsGuard warned that chatbot users are at risk of being fooled by foreign influence operations and other bad actors. The London-based Institute for Strategic Dialogue described the midterms as the first US election with AI tools built into every major tech platform.
Usage is already broad. In a survey by the National Conference on Citizenship, two in five Americans said they had used AI tools to search for election information, including details on candidates or ballot measures. Only 40% of respondents said they trust the tools with logistical information such as when, where and how to vote, and 60% said they trust them only a little or not at all to be unbiased about candidates.
The Brennan Center for Justice found a stronger record on conspiracy theories. Its August report tested ChatGPT, Gemini, Grok, Claude, Perplexity and DeepSeek, and every chatbot disputed every election conspiracy theory it was given, including claims that voting machines were rigged, even under repeated questioning. The same report found that half of all responses contained an inaccuracy or a bad citation.
The Brennan Center said that while the chatbots often made mistakes, they behaved as sympathetic listeners to skeptical users and still refused to endorse conspiracy theories. It concluded that chatbots "could play a constructive role in fighting election disinformation."
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
