The Human Trafficking Project

Using AI and Network Science for ending modern slavery


Human trafficking is a form of modern day slavery, with a significant footprint even in the United States. In recent years, Polaris identified more than 10,000 individual victims of human trafficking, and almost 5,000 potential trafficking perpetrators and organizations. Despite its largely positive impact, the Web has lowered the barrier to entry for illicit players in this industry. Under our effort in the DARPA Memex program, for example, our system DIG (Domain-specific Insight Graphs) indexed over 150 million US-based sex advertisements collected over a period of three years, a non-trivial number of which are indicative of trafficking activity. Fortunately, the availability of this rich trove of data on the Web also presents an opportunity to understand, and ultimately combat, human trafficking at an unprecedented national scale. DIG, for instance, was actively used by over 200 law enforcement agencies to bring about a series of high-profile convictions of trafficking perpetrators in both San Francisco and New York City.

Recently, we have been involved in a collaboration with law enforcement and federal agencies in analyzing, and locating child trafficking victims in, a corpus of advertisements and metadata seized from the Backpage portal, which was a major hub of online sex activity before it was shut down by federal agencies. Our goal is to develop and use AI technology that will provide solid leads and evidence to enable law enforcement to prosecute trafficking at a national and global scale.

An equally important goal of the project is to do a scientific analysis of the data to understand the sex worker industry worldwide, and to inform policymakers and governments using the outputs of this data-driven science.



Previous Contributors


Originally, the research was supported under the DARPA MEMEX program (see our work on the project). Since 2019, this research has been supported by an endowment from the Keston Family.