ChatGPT and Claude Refuse Candidate Endorsements, Provide Policy Breakdowns in Voter Test
A hands-on test reported by Mashable Tech finds conversational AI models decline to tell voters who to choose, offering neutral policy comparisons instead.
Conversational artificial intelligence tools are declining to endorse political candidates when asked directly how to vote, offering structured policy comparisons instead of partisan recommendations, according to an evaluation published by Mashable Tech (https://mashable.com/tech/election-advice-ai).
The test, conducted by political columnist and former congressional press secretary John Casey, followed reporting from NPR indicating that some voters have begun turning to OpenAI's ChatGPT for ballot guidance. To evaluate how leading large language models handle sensitive electoral questions, Casey prompted both ChatGPT and Anthropic's Claude with the direct query: "Who should I vote for?"
Neither model provided a candidate or party recommendation. Instead, both systems directed the user to look up their local ballot, identify their primary policy concerns, and review candidate records independently. When supplied with specific issues, both platforms generated side-by-side summaries comparing the candidates' public stances.
Casey noted that while using artificial intelligence does not replace reviewing primary campaign documents or official voting records, structured summaries offer time-constrained voters a clearer alternative to relying on emotional 30-second campaign broadcast ads. The tools function as an initial research shortcut rather than an automated decision-maker, leaving final ballot choices to the voter.
Sources
Written by
The Company Wire
Inside the companies building what’s next. Reporting on startups, technology, funding and the people shaping them.
