Machine learning links online criminal networks through writing patterns

A new study from arXiv (2605.04080) applies authorship attribution and machine learning to connect online criminal activities, such as human trafficking and illicit trade. Offenders often hide behind anonymous accounts and frequently change identities, making network mapping difficult. The research demonstrates that individuals maintain consistent patterns in writing and image presentation across advertisements, even when attempting anonymity. By analyzing these patterns across large datasets, the method links related accounts and identifies repeated behavior in illegal online markets. The study also proposes ethical guidelines for responsible use of such techniques.