Modeling Narrative Volatility in Periods of Political Instability: An AI-Enabled Analysis of Northern Ireland

Misinformation research has largely focused on identifying false content, often relying on platform-specific data and contested labeling practices. This article advances an alternative perspective by conceptualizing information disorder as a structural vulnerability arising from political instability and fragmented pub- lic narratives. Focusing on Northern Ireland as a post-conflict case, we integrate political event data with measures of media discourse structure to operational- ize narrative volatility—the degree to which news narratives become unstable over time. Using a combination of statistical, machine learning, and regime-based analyses, we show that narrative volatility varies independently of reporting volume and is shaped by both temporal persistence and crisis-like political con- ditions. An embedding-based semantic robustness check further confirms that periods of high narrative volatility correspond to genuine dispersion in media dis- course. Taken together, the findings support a structural account of information disorder in which vulnerability is rooted in political instability and discursive contestation rather than isolated content anomalies.