Deep Learning & AI: tech against trafficking
Counter-trafficking top 40
Written by Marqueze Kennedy
Child trafficking has been a problem for centuries. The technological advancements of the internet era, however, have brought with them an alarming growth of trafficking.
Traffickers now use technologies such as social media, untraceable cryptocurrency, and the dark web to conduct their criminal activities. Though these technologies are used by criminals to lure and exploit vulnerable children around the world, many innovative technologists have made efforts to fight child trafficking by building tools to catch and prosecute traffickers, and to protect vulnerable children.
The ways these technologies are being used range from anonymous and secure crime reporting to using AI to scan and scrape CSAM (Child Sexual Abuse Material) from the web at a level of speed and accuracy far beyond human capability. The range of these technologies is truly remarkable.
Anti-child trafficking technologies use different methods to address distinct aspects of the problem. Click to learn more about the technologies within each of our five categories:
2. Awareness
3. Reporting
Deep Learning & AI
This includes a variety of services and organizations diving deeper into the technologies capable of preventing human trafficking, such as chat bots simulating victims to gather information on buyers, mapping and clustering dark web sites, and using algorithms to identify bitcoin users based on the sex advertisements they click.
UC Berkeley Anti-Trafficking Algorithms
Trafficking Type: Sex
Region: US
Sector: NGO
In 2017, Rebecca Portnoff, a doctoral student at UC Berkeley, “developed two algorithms aimed to scan through online sex advertisements and find human trafficking circles.” Portnoff said the goal was to bring together the ads that were comprised by a single individual in order to help U.S. authorities focus on those who are being trafficked.
The first algorithm “links advertisements to a single writer using stylometry, which is the study of people’s writing styles.” The idea is that the writing style of one individual can be traced across a multitude of ads in order to identify which individuals are being trafficked. “The second algorithm allowed Portnoff’s research team to link Backpage advertisements to bitcoin transactions.” Through this method, they could then trace which advertisements were being bought using the same bitcoin account.
Freedom Signal and Project Intercept , by Seattle Against Slavery
Trafficking Type: Sex
Region: International
Sector: NGO
Freedom Signal is a platform consisting of six pillars used to combat sex trafficking online. The platform utilizes “online search result ads in response to sex buyer searches to stop buyers before they engage, social media ads targeting likely sex buyers online for deterrence messaging, chatbots that simulate a trafficking victim via text message to deter, delay, and gather data on sex buyers, victim outreach via text message to ads where sex is sold online, deterrence websites that illustrate the harms and consequences of buying trafficked sex, data analysis to increase the accuracy and effectiveness of buyer disruption.”
Project Intercept uses chatbots that pose as decoys on prostitution ads to gather data “about how, when, and where people are attempting to buy sex online.” The program uses language AI that was molded by “survivors of trafficking and prostitution” with the goal of confusing buyers and “decreasing the confidence necessary for buyers to follow through on their exploitation or violate the law.” This technology is currently being used nationally to fight sexual exploitation and is now moving into the international sphere.
Child Rescue Coalition
Trafficking Type: Child
Region: International
Sector: Government, Private
The Child Rescue Coalition is a collaborative among “child exploitation investigators, police officers, digital forensic experts, prosecutors, child welfare agencies and corporate and private philanthropy” that work together to solve the issue of child sex trafficking on all fronts. Their technology, which has been used by law enforcement in every state and in over 90 countries across the world, focuses on “identifying, monitoring, and ranking the online criminal behavior of child predators on peer-to-peer file-sharing and chat networks.” Some of the system highlights include “providing law enforcement with real-time investigative and prosecutorial assistance” as well as “analytics that target the offenders at greatest risk of presently abusing children.”
Hyperion Gray Production
Trafficking Type: Child
Region: US
Sector: Private
Hyperion Gray Production has a whole suite of products, one of which specializes in dark web investigations. This product, Avatar, helps law enforcement catch criminals on the dark web by shedding light on these hidden areas of the internet. In January 2018 they counted, mapped, and clustered 6,608 dark web sites. The map has 3,747 web sites as of March 2019 with embedded images disabled “in order to avoid the need to make redactions.” Users need to accept a disclaimer to view the map, which may include mature and/or offensive material.
Spotting Trends
Trafficking Type: Sex
Region: US
Sector: Private
In 2018, at Facebook’s annual child safety hackathon, multiple engineers from tech giants all around Silicon Valley came together to design new tools to fight child sexual exploitation. Spotting Trends designed a technology that uses “clustering analysis to keep tabs on traffickers” as they find new locations on the internet to conduct illegal activity. With large online prostitution marketplaces like Backpage being taken down, traffickers move to new areas of the internet which then need to be found.
SIGHT
Trafficking Type: Child
Region: US
Sector: NGO
Sovereign Intelligence Against Human Trafficking (SIGHT) uses deep machine learning technology to scrape petabytes worth of data from the dark web and turn that into actionable information usable by “victim service organizations, academic researchers, legal teams, and law enforcement agencies” to help them ensure they have the “best new data.” SIGHT is currently working to expand the scope and the nuance of a taxonomy of human trafficking entities and ontologies in online spaces.” They plan on using this data to better understand the different interconnected aspects of human trafficking and how it develops over time. The project’s goal is to use data to help law enforcement and policy makers make informed decisions in the fight against human trafficking.
Sweetie 2.0
Trafficking Type: Child
Region: International
Sector: Private
Sweetie is a 10-year-old virtual Filipino girl, created by Terre des Hommes, that has been used to unmask “over 1,000 offenders of webcam sex with children in just two months.” The project was created due to the booming industry of webcam child sex tourism where men seek children online and ask them to perform sexual acts through their webcams for money.
The initial campaign to raise awareness about this issue and research new ways to track these predators was massively successful. Government organizations around the world have already started using the data collected by the project to implement new regulations. Building upon their original approach, Terre des Hommes has begun working on Sweetie 2.0 which will include “a new advanced method to solve the problem of worldwide webcam child sex structurally.”
Sheepdog Bloodhound Internet Safety
Trafficking Type: Child
Region: International
Sector: Private
Sheepdog Bloodhound Internet Safety, founded in 2016, is an online organization that describes themselves as bloodhounds acting as sheepdogs, “protecting the sheep against the wolves.” Their goal is to make the internet a safer place by “searching for victims and identifying predators” and their crimes against children. Many of the posts also include AMBER alerts, online safety tips for parents, and tools that children and parents can use to help protect against predators online.
Microsoft Cognitive Services
Trafficking Type: General
Region: US
Sector: Private
During Ignite 2018, Microsoft announced a “$40 million, five year project” meant to help fund development in AI specifically geared towards humanitarian efforts around the world. Among these humanitarian efforts are “predictive analytics and bot frameworks” that bring a deeper understanding to the issue of human trafficking. Microsoft is already using its technology to combat this issue, but depending upon its success, more money may be funneled into this project to turn it into a major service.
This article is adapted from a 2019 white paper.
