Agentic AI on the Edge of Transparency: The OCCVI Observatory and the ARCON Pipeline for Mapping Crime, Corruption, and Harm in Latin America and Beyond

This paper presents a new epistemic framework for analyzing transnational corruption and organized crime using&nbsp;<i>Agentic Artificial Intelligence</i>(AAI). It introduces ARCON (Automated Robotics for Criminal Observation Networks) and OCCVI (Observatory of Crime, Corruption, and Victimization) as a semi-autonomous analytical system capable of transforming dispersed media data into structured knowledge about macro-criminal networks. The first section explores the limits of traditional investigative approaches in contexts of fragmentation and opacity. The second section conceptualizes corruption as a criminal phenomenon that thrives in the absence of transparency, arguing that AAI represents themost recent frontier in exposing its operational logic. Section three presents ARCON’s analytical cycle, from data ingestion to network graph generation through VORISOMA, culminating in OCCVI’s synthetic outputs. The fourth section introduces the&nbsp;<i>Structural Archeology of Harm</i>collection, which maps causal links between criminal structures and systemic victimization across Latin America. The final section reflects on the normative and methodological implications of integrating AAI into long-term anti-corruption strategies, emphasizing the need to center harm as an analytical category.