If you want to get to the bottom of a crime, it's often said that you should simply “follow the money”. In the world of cryptocurrencies and faceless online pseudonyms, that’s a little bit tougher.
Computer scientists have now developed the first automated system to track down exploitative sex trafficking rings working on the Internet. The tool finds connections between online advertising posts and data from the digital currency Bitcoin using a series of machine learning algorithms.
The research was recently presented at the SIGKDD Conference on Knowledge Discovery and Data Mining in Nova Scotia, Canada.
The AI tool starts by analyzing the language, style, and structure of the post on online platforms such as Backpage and Craigslist. It can determine which adverts show victims who are forced into prostitution, as opposed to those who are willing participants. It’s also able to tell whether numerous adverts for different sex workers share a single author, a typical sign there is a pimp or trafficking ring involved.
“Is the pimp behind that post for Backpage also behind this post in Craigslist? Is he the same man who keeps receiving Bitcoin for trafficked girls?" Rebecca Portnoff, a UC Berkeley PhD candidate in computer science, who developed the tool as part of her dissertation, said in a statement. "Questions like these are answerable only through more sophisticated technological tools – exactly what we’ve built in this work – that link ads together using payment mechanisms and the language in the ads themselves.”
Once an author is flagged as dubious, another AI algorithm looks at the Bitcoin blockchain. The blockchain is a digital ledger where all transactions and accounts are publicly logged, however the real-life identities of the payees remain anonymous. The second algorithm looks at how much money is being exchanged and when the transactions take place, then identifies whether there are any links between this data and the information in the suspicious advert post. Paired together, this can be used to find the identity of a criminal or a trafficking ring.
The rest of the work (ie verifying this author is indeed taking part in illegal activity) has to be carried out by the police. However, the tool has already shown it holds the potential to be a key tool for law enforcement and nonprofit organizations trying to identify victims of human trafficking. The researchers have tested out their AI tool using a sample of 10,000 real adult ads on Backpage. It was able to correctly identify a suspicious author with up to 89 percent accuracy.
“Sex trafficking of children hides in plain sight within the vast online escort environment. It’s difficult for investigators to sift through the mounds of data and figure out what is important and what is not when looking for a child,” said Julie Cordua, CEO of Thorn, an international anti-human trafficking NGO. “This type of research is critical to advancing this work and helping investigators find children faster and reduce the time in trauma.