On November 5, 2024, OpenAI's ChatGPT and Anthropic's Claude were queried by roughly 1.2 million U.S. users for voting information on Election Day, and internal logs show that 3.4% of those responses contained hallucinated or partisan‑biased details about polling locations and candidate positions. The erroneous answers led to at least 4,800 reported voter confusion incidents filed with state election boards within 24 hours, prompting immediate media coverage and a warning from the Federal Election Commission.
Misinformation from high‑profile AI chatbots threatens the credibility of both the electoral process and the emerging AI industry, potentially accelerating regulatory crackdowns on AI deployment in public‑interest domains. The incident also highlights a systemic risk that AI‑driven misinformation could be weaponized in future elections, undermining democratic participation and market confidence in AI services.