Using AI to Detect Human Trafficking from Space in the Brazilian Amazon

Powered by AI neural networks, the Human Trafficking Data Lab’s remote detection technology accurately pinpoints precise locations in and around the Brazilian Amazon where illegal deforestation and forced labor likely converge in producing charcoal used in the steel supply chain. The Charcoal Anti-Trafficking Reconnaissance (ChAR) project transforms a previously slow, reactive anti-trafficking response model into an efficient and proactive monitoring protocol that prioritizes the highest risk sites and the densest clusters of activity. 

Interact with a StoryMap about this project

Every year, an estimated 5,000 square miles of native Brazilian rainforest are cleared and converted into agricultural and cattle ranching land, a process intertwined with violence and exploitative labor. Identifying these critical intersections amplifies our Lab’s efforts to counter both labor exploitation and the environmental repercussions of tree loss and charcoal burning, thus marking a profound advancement towards social and environmental sustainability in the region.

Developed with low-cost satellite imagery and geospatial data and piloted in the state of Maranhão in collaboration with Brazilian authorities, the ChAR tool combines technical innovation with local expertise. Until recently, authorities relied on tips, many of which were anonymous, to locate charcoal camps suspected of using forced labor. Given the overwhelming number of tips and the significant challenge of prioritizing and determining the exact location of remote sites before traffickers moved operations to avoid detection, this technology analyzes satellite imagery to predict the locations of charcoal production sites linked to exploitation. This ensures a more efficient resource allocation, directing vital attention precisely where it's needed most. Based on pilot observation, ChAR makes substantial cuts in the time needed to carry out task forces and provides critical intelligence to assist in planning and safely executing site searches.

Using a quasi-experimental study design, we are measuring ChAR's impact on: (1) the probability that a known charcoal kiln is inspected; (2) the number of charcoal kilns inspected per task force; and (3) the number of workers rescued from charcoal production sites. Together, these measures evaluate whether the introduction of remote detection technology increases the responsiveness, efficiency, and impact of these anti-trafficking operations.

Because illicit charcoal is often produced from native forest in both the Amazon and the Cerrado - the most biodiverse savanna in the world, the ChAR tool represents a model for linking human rights protection with environmental sustainability. 

Location: State of Maranhão, Brazil

Funders

Partners