In the fight against human trafficking, a new digital tool developed at the University of California is giving investigators a powerful edge. By combining machine learning with bitcoin's public ledger, the system can sift through vast numbers of online sex advertisements to spot patterns that point to traffickers and their victims.
Rebecca Portnoff, a PhD candidate leading the project, explained that the tool examines details like the cost of an ad and its timestamp, then connects that ad to a specific person or group. This gives police a solid lead for further investigation, she told New Scientist.
How Bitcoin Became a Clue
The choice of bitcoin as a payment method on the website Backpage, long associated with sex trafficking, turned out to be a key vulnerability for criminals. In 2015, credit card companies stopped allowing their cards to be used on the site, pushing users toward bitcoin. Since bitcoin transactions are recorded on a public blockchain, the researchers can trace payments back to the people behind them.
This approach has already shown results. During a one-month trial, the tool accurately identified the same author for 90 percent of the 10,000 ads it reviewed. In one case, a single bitcoin wallet was used to pay for $150,000 worth of sex advertisements, hinting at a large-scale exploitation operation.
Scope of the Problem
The International Labour Organization estimates that nearly 4.5 million people worldwide have been forced into sexual exploitation. The new tool aims to help investigators intervene earlier, potentially preventing further harm to victims. Carrie Pemberton Ford of the Cambridge Centre for Applied Research in Human Trafficking noted that the tool could also assist in prosecuting suspected traffickers.
While the technology is not a silver bullet, it represents a practical step forward in using publicly available data to disrupt illegal networks. As the researchers refine the system, law enforcement agencies may gain an increasingly effective way to identify and act on trafficking patterns.

Comments
0