How AI can fight human trafficking

There are 40.3 million victims of human trafficking globally, according to the International Labor Organization. Marinus Analytics, a startup based in Pittsburgh, Pennsylvania, hopes to make a dent in that number. The company’s mission is to “serve those working on the frontlines of public safety by developing technology for them to disrupt human trafficking, child abuse, and cyber fraud.” For its achievements, Marinus won $500,000 as part of its third-place ranking in the 2021 IBM Watson AI XPRIZE competition. The startup is the brainchild of three co-founders: Cara Jones, Emily Kennedy, and Artur Dubrawski, who launched it out of the Robotics Institute at Carnegie Mellon University in 2014.

Marinus implements its mission primarily through its set of AI-based tools called Traffic Jam, whose goal is “to find missing persons, stop human trafficking and fight organized crime.”

How AI helps

Traditionally, finding a missing person would involve taping a picture of the person on the computer and then manually combing through thousands, if not millions, of online ads on adult services websites to see if any of the posted pictures match. Such a process is time-consuming and tiring. A human detective’s attention can start flagging after long hours at the computer doing the same task endlessly.

AI takes on this sifting task effectively. It combs through millions of online ads in specialty “hot spots” (commercial adult services websites) and searches for “vulnerability indicators.” Cara Jones, cofounder, and CEO says vulnerability indicators can mean a number of things that act as red flags. These can include images of subjects who look like children and indications of drug use.

AI is especially useful in handling unstructured data, which is not presented in familiar digital formats. Computer vision also helps analyze imagery and scan for potential problems. Importantly, AI can do such work at scale.

Because Traffic Jam helps with the manual labor involved in sifting through millions of online ads, detectives can divert their limited resources to more pressing issues like tracking down victims and dismantling networks. Traffic Jam narrows the field of search and does not make definitive pronouncements on crimes. “We reveal intelligence quickly and save time so crime-fighters can react faster. There is a strong element of human review throughout the process,” Jones said. “It is not a black box.” Traffic Jam is transparent in explaining the connections that are being made.

The AI can observe patterns of trafficking among the data and the larger criminal networks at play. “Law enforcement is very jurisdiction-based, and they’re usually only able to see what’s happening locally,” Jones pointed out. “Some of this activity is happening on a transnational level, so we can connect the dots and reveal the larger systemic trafficking that’s happening and inform law enforcement.” In 2019 alone, Marinus Analytics identified 3,800 victims of trafficking.

Crime-fighting agencies are increasingly realizing the value of digital tools like Traffic Jam. As Jones explains, “The agencies recognize that they need purpose-built capabilities, and we have been able to show what’s possible.” Marinus claims that in 2020 alone, Traffic Jam saved 70,000 investigative hours.

Marinus’ next steps involve moving into the public safety and human services spaces, such as child protection.

“[Trafficking] is not an amateur operation. There are very sophisticated establishments that have a value chain to conduct their business in these spaces, and the more we can start to reveal the business practices to the right law enforcement audience, the better,” Jones said. “We are in a new era of public safety response.”