AI chatbots used by voters face scrutiny over election accuracy

Voters are increasingly turning to AI chatbots to research ballots, compare candidates, and verify claims, particularly for down-ballot races that receive less media coverage. While some users find these tools helpful for organizing information, experts warn that the technology can be subtly persuasive or reinforce existing biases without providing new information.

A report by the California Initiative for Technology and Democracy (CITED), a project of Common Cause, highlights significant risks regarding accuracy. The analysis found that major chatbots, including Anthropic’s Claude, OpenAI’s ChatGPT, and DeepSeek, produced errors regarding specific election details. Examples include incorrect information about Los Angeles mayoral race outcomes and directing users to outdated or incorrect ballot propositions in California.

Because AI companies do not disclose how their models are trained or which sources they prioritize, concerns remain regarding the potential for chatbots to mislead voters through factual errors or inherent model bias.