AI chatbots gave voters incorrect information in voting guides, Washington Post finds
The Washington Post tested more than 3,000 AI-generated voting guides and found frequent errors, including incorrect congressional districts. A late-October Pew survey found 8% of U.S. adults had used a chatbot for 2026 election information.
Key points
- When chatbots named a district, they got it wrong 22% of the time
- The guides included errors such as a nonexistent Missouri Senate race and misidentified Wyoming candidates
- Pew found 8% of U.S. adults had used chatbots for 2026 election information
- Amazon paused Alexa’s voter-guide feature; Microsoft said it was investigating
Sources
- The Japan TimesRead ↗Spewing U.S. election falsehoods, AI chatbots face scrutiny
- applescoop.orgRead ↗AI voter guides got the ballot wrong: check yours here
- FuturismRead ↗ChatGPT Is Giving Voters Ludicrous Misinformation Right as the Midterms Hit
- Gadget ReviewRead ↗ChatGPT Is Giving Voters Serious Misinformation as the Midterms Hit: Fake Candidates and Wrong Election Information
- The New York TimesRead ↗Tech Giants Try to Curb A.I.-Generated Misinformation About the Midterm Elections
Details
The Washington Post asked six popular AI assistants to create midterm voting guides for addresses across U.S. counties. It found that the guides often misstated which congressional district an address was in, and some included races or candidates that did not belong on the voter’s ballot. The Post reported that Claude and Alexa+ produced incorrect information most often; ChatGPT, Google AI Mode and Perplexity more often failed to name a district at all.
Examples in the Post’s analysis included ChatGPT adding a Missouri Senate race that was not on the 2026 ballot and Claude presenting Wyoming candidates for state offices as Senate contenders. The Post also found chatbots citing voter-guide websites that appeared to use AI-generated content. Microsoft Copilot cited one such site, Ballot Brief, in 75% of its guide answers, while ChatGPT cited it in 22%.
Companies described different responses. Amazon said it had identified a cause and paused Alexa’s ability to create voter guides, while Microsoft said it was looking into the results. OpenAI and Anthropic pointed to partnerships with Democracy Works, and Google said AI Mode checks generated claims against trusted data. The Post reported that OpenAI and Amazon appeared to change some responses after being contacted.
A Pew Research Center survey of 5,273 U.S. adults, conducted October 12–18, found that 8% had used an AI chatbot for 2026 election information. Among those users, 40% sought logistical information such as polling places or mail-ballot rules.
Written from coverage as of
Images
- applescoop.org
- Futurism
- Gadget Review
Stocks in this story
Change since open: 20 Nov to 21 Nov 2026
Last refresh