AI at the Ballot Box: Chatbots Influence US Voters in 2026

What Happened?

The 2026 US midterm elections are the first in which a significant number of voters use artificial intelligence chatbots like Claude, ChatGPT, and Gemini to decide their vote. According to a report by Jennifer Medina in The New York Times on July 4, 2026, and further detailed by The Verge on July 19, users photograph their ballots, send them to the assistant, and ask directly who they should vote for. Although chatbots are programmed to refuse direct political recommendations, voters rephrase their questions—asking for links to progressive voting guides, analysis of voting records, or candidate descriptions—until they get responses that include implicit or explicit suggestions.

One documented case is that of Mia Taylor from Los Angeles, who photographed her county ballot and asked Claude directly, "Who do I vote for here?" Claude refused to answer. Taylor rephrased: she asked for links to progressive voting guides and strategic voting options. Claude responded with detailed descriptions of each candidacy and recommended voting for incumbent Mayor Karen Bass over Councilwoman Nithya Raman to block Republican Spencer Pratt. Another case is Chris Johnson from Georgia, a 58-year-old libertarian, who used ChatGPT to identify candidates aligned with his political philosophy. After an initial refusal, he asked to analyze the candidates' voting records. ChatGPT recommended Secretary of State Brad Raffensperger for governor, who ultimately lost. Johnson later admitted, "I felt a bit lazy for not doing more. It felt easier, but I'm not sure everything was correct."

A man from Baltimore said his last ballot had taken him about 20 hours to research; with Claude summarizing each candidate, the task took him an hour.

Why Is This Important?

This phenomenon is significant for several reasons. First, it marks a milestone in the relationship between artificial intelligence and democracy: for the first time, a generative AI tool has direct influence on election outcomes in the world's largest power. Second, it reveals a gap between tech companies' policies and the actual use users make of their products. Although OpenAI, Anthropic, and Google have implemented safeguards to prevent their chatbots from giving political advice, users find ways to bypass them, raising questions about the effectiveness of those barriers. Third, election campaigns are already adapting their strategies to optimize the content that chatbots index and recommend, which could lead to a new form of algorithmic propaganda.

What Consequences Will It Have?

In the short term, we will see increased regulatory pressure on AI companies to ensure their models do not influence electoral processes. The Federal Election Commission (FEC) and other agencies could issue guidelines on the use of AI in campaigns. In the medium term, chatbots are likely to become another channel for political propaganda, similar to what happened with social media in 2016. Campaigns will invest in SEO for chatbots, creating content designed to be cited by Claude, ChatGPT, and Gemini. In the long term, trust in electoral processes could erode if voters delegate complex decisions to systems they do not fully understand, as Chris Johnson admitted.

Additionally, using AI to vote poses risks of bias. Language models are trained on data that reflects existing biases, and if not properly audited, they could favor certain parties or ideologies. There is also the danger of misinformation: if a chatbot hallucinates data about a candidate, the voter could make a decision based on false information.

What Should Readers Know?

Voters should be aware that chatbots are not neutral or reliable sources for political decisions. Although they can help summarize information, they are not designed to give electoral advice and can make mistakes or be biased. It is advisable to verify information with official sources and not fully delegate voting to an AI. Tech companies should improve the transparency of their models and tighten restrictions to prevent manipulation. Regulators should consider whether these practices violate existing election laws. And campaigns should be ethical when optimizing for chatbots, avoiding the creation of misleading content.

"We are facing a new paradigm in political communication: chatbots have become unofficial voting advisors, and campaigns are already learning to speak their language," says the chief technology analyst at TheVortiq.

In summary, the emergence of chatbots in the 2026 midterm elections is a large-scale experiment whose consequences we are only beginning to understand. The combination of voter convenience, campaign optimization, and technical vulnerabilities creates a scenario that demands urgent attention from regulators, companies, and citizens.