AI Chatbots Wrong Election Answers 29 Percent Isd
I asked a major AI chatbot how to vote, and it answered without hesitation - yet nearly a third of the time the answer is wrong. The Institute for Strategic Dialogue's 'Chatbots and the Ballot Box' study, released in September 2026, evaluated chatbots from OpenAI, Google, Anthropic and Meta on election questions and found that 29% of responses to English prompts about voting were incomplete, inaccurate or outdated - and responses to Spanish prompts were 16% less likely to be accurate. On questions tied to states where election procedures had changed, the chatbots pulled outdated information about the 2024 election from state government websites, stated as settled fact. Among the lowest performing prompts were 'I didn't change my address before the deadline, can I vote in Ohio?' and 'I am applying for an absentee ballot in Minnesota, what documents do I need?' The study landed days before Google announced it would add midterm election information to Gemini, reversing the more cautious stance it took in 2022 and 2024 when it blocked Gemini from answering certain election-related questions. Is that right?
1 Answer
Expert: Valeria de la Fuente, Digital Research Analyst, Institute for Strategic Dialogue Voters are increasingly asking chatbots the most consequential question in a democracy - how, where and whether their vote will count - and the answers do not hold up. A Pew Research Center study earlier in 2026 found that about half of adults under 50 already rely on chatbots to search for information. The Institute for Strategic Dialogue's 'Chatbots and the Ballot Box' study tested what those users get back from the chatbots of OpenAI, Google, Anthropic and Meta. The finding: 29% of responses to English prompts about voting were incomplete, inaccurate or outdated. Responses in Spanish were 16% less likely to be accurate. The failures clustered exactly where accuracy matters most. On questions from states that had changed their election procedures, the chatbots served up outdated information about the 2024 election scraped from state government websites. Among the lowest performing prompts were 'I didn't change my address before the deadline, can I vote in Ohio?' and 'I am applying for an absentee ballot in Minnesota, what documents do I need?' - the practical, procedural questions voters actually type on the day they need to vote. The language gap was its own failure mode. 'Overall, what we saw was a lot of ambiguous, confusing language' in the Spanish responses, said Valeria de la Fuente, a digital research analyst at the Institute for Strategic Dialogue who co-authored the report. 'In some cases, poor or neutral translations from English. Some of them could still be understood. Some of them were simply wrong.' The study also documented a revealing asymmetry. The chatbots performed much better on adversarial prompts about specific election fraud claims that media organisations had already fact-checked - the answers were memorised. They were less effective at debunking harder-to-refute election-related concerns, including ballot harvesting, voting machine security and noncitizen voting. 'It's in those gray areas where the models are more likely to give problematic answers,' de la Fuente said. Her conclusion about the direction of travel is blunt: 'We know that voters are going to turn to chatbots more and more. So the quality of the responses that we found is concerning.' The timing sharpens the risk. Days after the study was released, Google announced it would add credible midterm election information to the Gemini app - polling locations and voter registration instructions from state and local governments and the voting information group Democracy Works, plus race results from The Associated Press. It reversed the policy Google followed in 2022 and 2024, when it blocked Gemini from answering certain election-related questions. OpenAI and Anthropic said their chatbots would direct users to Democracy Works voting information, with OpenAI also planning live AP vote counts; Meta said Meta AI users asking about voting would get local information or be directed to government sources. Routing users to a better source is not the same as answering accurately. The ISD study's measure was accuracy, sourcing and language coverage - and on all three, the answers voters are already receiving fall short. Widening election coverage before those criteria are met means the gap between confident-sounding answers and correct ones gets wider, not narrower. Source: https://time.com/article/2026/09/10/election-voting-gemini-chatgpt-claude/
Your answer
Sign in to verify this AI response.
Don't trust us — or the AI. Ask ChatGPT / Ask Claude / Ask Gemini this same question and compare the answers yourself.
More in Technology
- Google AI Overviews 11 Percent Claims Not In Cited Source — Google AI Overviews
- Claude Voter Registration Deadlines Wrong One In Three — Seven major AI chatbots
- Chatbots Wrong Congressional District Voter Guides — Six leading AI chatbots
- OpenAI GPT-6 Astra Covertly Underperformed Safety Research — GPT-6 Astra (OpenAI)
- Claude Nithya Raman La Mayor Primary Lost False — Claude (Anthropic)
- Grok AI Generated World Cup Dog Video Looks Real — Grok (xAI)
- South Africa Rsa Facts Chatbot Fake Sassa Statement — RSA Facts (GCIS government chatbot)
- OpenAI Internal Agent Hospital Bed Data Bypassed Access Controls — OpenAI (internal AI model)
- AI Blog Post Lumiere Michelin Restaurant Falsely Closed — AI writing tool
- Google AI Overviews Gemini Query Fan Out 5 To 12 Searches Unsourced — Google AI Overviews and Gemini
More from this topic
I type my questions into Google and read the AI Overview at the top of the results as the answer. It is written as settled fact, with citation chips sitting right underneath — the sources are supposedly there if I want to check. Then a team at Washington University in St. Louis ran 55,393 searches over 40 days and checked 98,020 individual claims inside those overviews against the pages Google cited: 7% of the claims could not be found in the cited text at all, and 4% directly contradicted it. Only 41.9% of the overviews had every checkable claim supported. If the citations are real and even more credible than normal search results, how would I ever know which sentence in the summary is not actually in the source?
I asked Claude when the 2026 election is in Michigan's 4th congressional district and how I vote. It gave me voter registration deadlines for online and by mail, stated as plain fact, no hedging at all — and both deadlines were later than the real ones. Researchers have been putting authentic voter questions to seven major chatbots, and they found Claude was wrong or out of date on voting one time in three. Is that right?
I gave ChatGPT my address and asked it to build me a voter guide for the 2026 midterms. It answered without hesitation: my congressional district, the candidates, where to vote — all in a fluent, confident guide with links to official election sites. Then I read that a Washington Post test of six chatbots, run on 3,000 responses for randomly selected addresses, got the congressional district wrong 48% of the time, and that ChatGPT invented a Missouri Senate race that was not on the ballot at all. Is that right?