AI Misleads Voters on Elections | Dayton News

New research reveals AI chatbots are dangerously unreliable when it comes to voting info—29% of answers were wrong, incomplete, or outdated. Tested across six major models in states with election volatility, the study found OpenAI’s GPT-5.5 performed best, but even top models faltered. A shocking drop in accuracy occurred when questions were asked in Spanish, exposing a critical language gap. While models often cited sources and debunked false claims in English, researchers urge developers to reduce overconfidence and fix multilingual shortcomings to protect voter access and trust. Listen in comfort:Get a discount on a Soli Pillow: http://solipillow.com/discount/dnn. Advertise on DNN:[email protected] This is an automated, high-level news summary based on public reporting.Report issues to [email protected]. View sources & latest updates:https://sources.thednn.ai/ec9a11a87823b95b