Efficient Disruption of Criminal Networks through Multi-Objective Genetic Algorithms

A team of researchers has published a novel AI framework that uses multi-objective genetic algorithms to optimize the disruption of real-world criminal networks like the Sicilian Mafia. The study, accepted for the 2026 IEEE Conference on Artificial Intelligence, moves beyond traditional Social Network Analysis (SNA) that focuses solely on removing high-centrality "key players." Instead, it introduces a practical constraint often overlooked in academia: operational costs. The framework treats law enforcement action as an optimization problem with two conflicting goals—maximizing network fragmentation while minimizing the spatial distance officers must travel from their headquarters.

The researchers implemented two algorithms—Weighted Sum Genetic Algorithm (WS-GA) and the more advanced Non-dominated Sorting Genetic Algorithm II (NSGA-II)—to find optimal disruption strategies. Using the real "Montagna Operation" dataset, they demonstrated that while centrality-based methods (like targeting the most connected individuals) can fragment networks, they often require logistically expensive, long-distance operations. The new AI approach finds alternative target sets that achieve similar disruption levels but with dramatically reduced operational burdens, making the strategies more feasible for real Law Enforcement Agencies (LEAs) with limited resources.