Synthetic Politics: Preparing Democracy for Generative AI

Published ahead of the 2024 UK general election, Synthetic Politics provides a pre-election assessment of generative AI risks to democratic processes — covering synthetic media, personalised disinformation, autonomous political agents, and AI-enabled voter suppression. The breadth of the risk taxonomy is the report's most useful contribution: it maps a wider range of AI-democracy threats than most pre-election assessments, which tend to focus narrowly on deepfakes.

The autonomous political agents dimension is particularly forward-looking and underweighted in the current literature. As AI systems become capable of conducting persistent, adaptive political communication at scale — generating tailored messages for individual voters, operating social media accounts, and responding dynamically to political developments — they create a category of influence operation that is qualitatively different from any existing threat. The policy framework the report proposes — platform regulation, election commission oversight, and prebunking — is calibrated to the threat landscape as of 2024; the autonomous agents threat will require more fundamental responses as the technology develops.