Stanford Researchers Investigate Human Trafficking Alongside Brazilian Partners in the Amazon Rainforest

Other core members of the Stanford Human Trafficking Data Lab are Luis Fabiano de Assis, PhD, an affiliate scholar at the Center for Human Rights and International Justice; Mike Baiocchi, PhD, an associate professor of epidemiology and population health at Stanford Medicine; Kimberly Babiarz, PhD, a social science research scholar at Stanford Health Policy; Victoria Ward, a clinical associate professor of pediatrics at Stanford Medicine; Jessie Brunner, MS, associate director of strategy and program development at the Center for Human Rights and International Justice; and Lizzy Constantz, the lab’s program manager. 

Image

Data Lab members from left: Luis De Assis, Thay Graciano, Kim Babiarz, Sierra Wells, Dayana Coelho and Sofia Penglade.

The Stanford trafficking lab is conducting research in Brazil and expanding a comprehensive anti-trafficking database, in part with a $1.3 million grant from the U.S. State Department’s Office to Monitor and Combat Trafficking in Persons. Their key partner in Latin America’s largest country is Assis, a federal prosecutor, computer scientist, and the head of the SmartLab initiative at the Federal Labor Prosecution Office (FLPO), where he blends his legal and data science background.

“An accurate AI-driven remote detection technology can be a game changer in combating trafficking,” said Assis. “Given the overwhelming number of tips and the significant challenge of prioritization, this technology can ensure a more efficient resource allocation, directing vital attention precisely where it's needed most.”

The Stanford lab and Assis worked together to develop software that detected 21 likely charcoal production sites to target over the summer, sites that were further investigated by government partners before an anti-trafficking task force was formed to inspect the sites in person.

Brazil has been lauded for taking serious steps to end modern slavery within its borders. Its public registryknown as the “Dirty List,” targets individuals and companies suspected of human trafficking. Those on the list face financial penalties and are prohibited from obtaining credit through public banks—and suffer reputational harms.

Field Research

Miller, a professor of health policy and senior fellow at the Freeman Spogli Institute for International Studies and the Stanford Institute for Economic Policy Research, recently visited several of these charcoal kiln camps in the northeastern state of Maranhão, parts of which sit in the Amazon Rainforest. He traveled with several other Stanford researchers, inspectors, and prosecutors escorted by armed officers of the Federal Police. This multi-agency team is charged with finding the perpetrators and victims of trafficking. 

There are thousands of acres of these farms dotted with mud mortar kilns that look like camel-colored igloos. Men feed native trees into the kilns, producing charcoal which is then used to make pig iron—a primary source for steel—which is then sold globally to construction, shipbuilding, aerospace and automotive industries, to name a few. New technology being developed can help foster accountability within these supply chains, emphasizing the necessity for businesses to ensure ethical practices throughout their operations. 

One of the federal prosecutors traveling with Miller had a PM 2.5 monitor which measures fine particulate matter that can be inhaled into lungs. They found the workers were exposed to a magnitude 200 times worse than a healthy level.

“And the closer you get to these ovens, you see a topcoat of soot that quickly becomes unhealthy by magnitudes of thousands,” Miller said. “The conditions are pretty horrific.”

Miller noted the coal ovens are sealed to cook the biomass for a week or two and must be constantly monitored so the charcoal doesn’t get too hot or too cold. He interviewed some of the workers who explained the process of keeping the fires burning night and day. One man, who initially declined to give his name, was asked if he wanted to go with the officers who would help him resettle, but he would not commit nor say whether he was working the ovens against his will. He said that he worked 16 to 18 hours a day and slept lightly during his off hours as he had to constantly check his ovens.

“In some of these situations, the dynamics are subtle and even the folks that we determine to have been trafficked may have complicated feelings about being removed from the situation,” said Baiocchi, who helped design the algorithms used to detect the charcoal operations. 

He traveled with the lab team to see the charcoal farms firsthand. “The work these Brazilian inspectors and prosecutors do is really important and complex; they have to make quick decisions on the spot to prevent leaving trafficked people behind,” he said.

Miller emphasized that he and his team are not working to “rescue” laborers who might be trafficked, but to learn more about their health and welfare conditions and work with the Brazilian government on identifying the illegal logging and coal oven sites.

“You know we’re all committed to the cause behind our work—to help those who have been trafficked and then focus on their recoveries—but seeing those conditions, it brings tears to your eyes,” Miller said. “To be part of that team that is trying to contribute in some small way to a solution is more meaningful than most of the other things I’ve done as a professor.” 

Tracking the Traffickers

Until now, identifying charcoal camps believed to be using forced labor has been reliant on tips, many of which are anonymous. With an overwhelming number of tips to process simultaneously, prioritizing which leads to follow becomes a challenge for the authorities. 

“Tracking down the exact location of these sites can be extremely challenging, and it used to take days to pinpoint them,” Assis said. And even once a camp is identified, the traffickers frequently pack up and move the workers to avoid detection. Consequently, anti-trafficking agents are often too late to catch the traffickers in the act and help victims.

So the Stanford trafficking lab members, working with Brazilian agencies and officials, have been developing a machine-learning tool they call CHAR—charcoal anti-trafficking reconnaissance. It uses satellite imagery from PlanetScope and then feeds these into the algorithm trained to take that data and predict the locations of charcoal producing sites. 

The data hub offers an unprecedented glimpse into the complex world of trafficking in Brazil, uniting thousands of real trafficking cases with intricate details that shed light on the challenges survivors face and the tactics used by perpetrators.

"It's very exciting to bring these powerful new technologies to the fight against human trafficking,” said Babiarz, a social science research scholar at Stanford Health Policy who is an expert on large-scale impact evaluations, econometrics and machine-learning analysis. She is the senior analyst for the lab’s research initiatives. 

“This summer, we saw first-hand how the tools we have been developing can bring value to the tremendous work of frontline anti-trafficking agents in Brazil, making a real difference in the lives of exploited workers,” Babiarz said.

Brunner said that when she began researching labor exploitation a decade ago, the lack of available data inhibited a robust understanding of the scale and scope of human trafficking around the globe — in which 50 million people are believed to be trafficked into modern slavery on any given day.

“But our Lab is demonstrating that administrative data combined with innovative technologies and data science methods can shed light on this terrible human rights violation with the aim of real-world impact,” Brunner said. “And I think that connects us directly to Stanford's mission as a purpose-driven university committed to benefitting society.”

Other supporters of projects underway at the trafficking data lab include the Stanford King Center for Global Development, the Stanford Center for Innovation in Global Health, the Stanford Human-Centered Artificial Intelligence center, Stanford Woods Institute for the Environment, and the Dolby Family.