Generative AI has prompted widespread concern about its potential effects on elections, from flooding the information environment with misinformation to enabling highly personalised persuasion and undermining trust in what is real. This talk critically examines six prominent claims about AI’s electoral harms against existing evidence from political science, communication, and related fields. I argue that while generative AI introduces genuine risks, many predictions of transformative effects on elections overlook persistent constraints such as people’s limited attention, the credibility of sources, people’s existing beliefs, as well as their patterns of information consumption. I conclude by arguing that an excessive focus on AI-driven threats may distract from more fundamental challenges to electoral integrity and democracy.

SBRB Living with AI – Webinar 2 – AI Misinformation and Elections – F Simon