Synthetic Persuasion: An Ethical Analysis of AI-Generated Political Content on Social Media

Generative AI has made it cheap and quick to produce persuasive political images, audio, and video. The conventional view is that the danger comes from deception: Voters will be fooled by the synthetic products and will change their minds about how to vote. In this article, I argue that the actual moral issue here is a deeper one. By using contemporary empirical research and combining deontological, consequentialist, and virtue ethics approaches to moral philosophy, I analyze what counts as misconduct when it comes to the production and distribution of synthetic political media created by AI. Deception proves to be rather uncommon, while inexpensive and low-tech fake content seems to dominate the market over sophisticated deepfakes; however, the synthesis of political media still threatens the viability of democratic governance. It affects voters subversively by bypassing their ability to reason, destroys the mutual trust necessary for public deliberation, and provides a liar with a "liar's dividend" – a possibility to discredit all evidence presented as a fabrication. All three philosophical traditions agree that these consequences are unacceptable, while allowing for legitimate, openly admitted, satirical, and educative use of the technology.