Connecting online criminal behavior with machine learning: Using authorship attribution to analyze and link potential online traffickers
New research uses writing patterns to unmask criminal networks hiding behind fake accounts.
A new doctoral thesis by Vageesh Kumar Saxena, published on arXiv (cs.CL), demonstrates how machine learning can be used to analyze and link online criminal behavior through authorship attribution. The research focuses on illegal activities like human trafficking and illicit trade that have migrated to online platforms, where offenders use anonymous accounts and frequently change identities to evade detection. By examining large collections of online advertisements, the study shows that individuals maintain consistent writing styles and image presentation patterns even when attempting to remain anonymous. These stylistic fingerprints allow the algorithm to connect related accounts across different illegal markets, revealing the scale and structure of hidden criminal networks.
The work goes beyond technical methodology to address responsible deployment. Saxena proposes clear guidelines to ensure privacy, fairness, and transparency when such tools are used by law enforcement. The research spans multiple fields including computation and language, artificial intelligence, computer vision, and social information networks. This interdisciplinary approach provides practical support for investigations while emphasizing ethical safeguards. The thesis, submitted on April 14, 2026, is available as arXiv:2605.04080 and related to a DOI linked to a 2025 dissertation.
- Uses authorship attribution to detect consistent writing and image patterns in anonymous online ads
- Links related accounts across illegal markets to map criminal network structures
- Proposes ethical guidelines for privacy, fairness, and transparency in law enforcement use
Why It Matters
ML can now unmask anonymous criminal networks online, but must be deployed with ethical guardrails.