Paper Type

Complete

Abstract

Political deepfakes have become a growing concern for democratic discourse, yet existing Information Systems (IS) research has focused primarily on individual-level perception and detection, leaving the question of how political deepfake narratives are distributed across social media platforms underexplored. Drawing on affordance theory, this paper conceptualizes social media platforms not as neutral hosts but as sociotechnical environments whose material features shape which synthetic narratives can be produced, amplified, and rewarded. Using the Political Deepfakes Incidents Database (PDID), we empirically analyze 1,498 political deepfake incidents across Facebook, Instagram, Reddit, TikTok, Twitter/X, and YouTube along three narrative dimensions: depicted topic, communication goal, and core rhetorical frame. Logistic regressions with robust standard errors and joint Wald tests show that depicted topics and communication goals vary significantly across platforms, whereas core rhetorical frames remain uniform. We interpret this divergence as evidence of a "two-layer" structure: platform affordances differentiate the surface-level narrative while leaving an invariant rhetorical grammar intact. The paper contributes a platform-level characterization of political deepfakes and offers a descriptive foundation for affordance-sensitive mitigation and media literacy.

Paper Number

1862

Comments

SIG ADIT

Download

Share

COinS
 
Aug 15th, 12:00 AM

Same Frame, Different Story: A Cross-Platform Analysis of Political Deepfakes

Political deepfakes have become a growing concern for democratic discourse, yet existing Information Systems (IS) research has focused primarily on individual-level perception and detection, leaving the question of how political deepfake narratives are distributed across social media platforms underexplored. Drawing on affordance theory, this paper conceptualizes social media platforms not as neutral hosts but as sociotechnical environments whose material features shape which synthetic narratives can be produced, amplified, and rewarded. Using the Political Deepfakes Incidents Database (PDID), we empirically analyze 1,498 political deepfake incidents across Facebook, Instagram, Reddit, TikTok, Twitter/X, and YouTube along three narrative dimensions: depicted topic, communication goal, and core rhetorical frame. Logistic regressions with robust standard errors and joint Wald tests show that depicted topics and communication goals vary significantly across platforms, whereas core rhetorical frames remain uniform. We interpret this divergence as evidence of a "two-layer" structure: platform affordances differentiate the surface-level narrative while leaving an invariant rhetorical grammar intact. The paper contributes a platform-level characterization of political deepfakes and offers a descriptive foundation for affordance-sensitive mitigation and media literacy.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.