Bitcoin mempool scan helps victims of human trafficking
TechnologyBit News Today1 min read
Computer science researchers at the University of California, Berkeley, have created new tools to help law enforcement detect and prosecute sex trafficking rings. These automated techniques find connections by analyzing public Bitcoin information corresponding to adult ads on sites such as Backpage and Craigslist.
The first study of its kind, UC Berkeley computer science Ph.D. candidate Rebecca Portnoff worked with other Berkeley researchers, as well as academics from UC San Diego and NYU Tandon, to develop this technology. Portnoff developed the technology as part of her doctoral dissertation and presented the paper at KDD 2017. The research is aimed not just at assisting law enforcement, but also non-profit groups striving to identify trafficking victims.
The system operates by comparing the text of ads to identify those from a common author, then tracing the Bitcoin transactions used to pay for them. Bitcoin became the preferred payment method after major credit card processors, such as Visa, refused service to sites like Backpage in 2015. Because the Bitcoin blockchain allows public tracking of transactions, the UC Berkeley team could link transactions to author, discovering that one Bitcoin wallet had paid for $150,000 worth of sex advertisements, suggesting a massive exploitation operation. Analyzing 10,000 real Backpage ads over four weeks, the researchers achieved an 89 percent true-positive rate in grouping them by author based on stylistic similarities. Following encouraging trials, law enforcement agencies, corporations, and non-profit groups have shown keen interest in the methods, with several already using or preparing to implement them.
Law enforcement could deploy these new algorithms to target sex trafficking organizations to identify victims and prosecute owners. The linkages might enable investigators to subpoena further records from the Bitcoin wallet services or expand on sting operations using phone numbers listed in the ads. The SIGKDD conference will include the paper in its published proceedings.



