Human trafficking is a massive global enterprise that generates an estimated $150 billion in annual profits. Human trafficking refers to a complex global phenomenon that includes sexual exploitation, forced labour, forced criminality, the deployment of child soldiers, and indentured servitude. It poses a constant threat to vulnerable women, children, and men worldwide. Each form of human trafficking can be attributed to variations in the larger socioeconomic system, a smattering of seemingly random instances that create vulnerabilities on one end and opportunities for exploitation on the other. Currently, the tools that are used to detect and potentially end human trafficking are far too limited in their utility because of just how complex a phenomenon human trafficking is. One potential solution to this problem is to deploy artificial intelligence (AI) and machine learning (ML). AI/ML is a scalable tool that enables computers to learn from themselves as they analyse data and recognise patterns, essentially erasing the limitation of the human programmer. It is capable of analyzing not only financial activity, but social, economic, and even political conditions from hundreds of thousands of data sources. AI/ML could minimize the complexity of human trafficking and vastly improve detection and prevention.
According to the International Labor Organization, there are40.3 million victims of human trafficking globally and most countries serve as a source, transit, or destination country for trafficked victims. Of the 40.3 million human victims, 16 million are forced into labor in the private sector. They can be found working on the supply chains of large multinational corporations, particularly in the fishing, textile, construction, mineral, and agricultural industries. The massive population of human trafficking victims exploited within the private sector inspired the UK to pass the Modern Slavery Act of 2015. This law requires all organizations with worldwide revenues of at least £36 million that operate in the UK to publish an annual transparency statement describing the efforts they took in the previous fiscal year to ensure that their business and supply chains are free from modern slavery and human trafficking.
The UK’s Modern Slavery Act underscores the massive economic impact that human trafficking has on the world economy, though some activists argue that the law’s impact has been and will continue to be minimal, considering it only requires a description of efforts and not an augmentation of those efforts. At the same time, many argue that since the law was enacted, there has been increased internal dialogue and awareness within these large companies.
In 2018, the courts in the UK used the Modern Slavery Act for the first time to convict someone for slavery offences and perverting the course of justice. Josephine Iyamu, a British nurse, was found guilty of slavery offences against five Nigerian women who were promised a better life in Europe and then forced to work as sex slaves in Germany. The number of recorded modern slavery crimes between 2015-2016 and 2016-2017 increased from 870 to 2,255; however, the conviction rate is believed to have decreased from 70% to 61%, which reflects the steep learning curve that law enforcement representatives face as they figure out how to implement the law effectively.
While the UK’s Modern Slavery Act is a step in the right direction, the data that has emerged since its inception highlights the need for a modern, holistic, and multidisciplinary approach to efforts tasked with reducing and ending human trafficking like AI/ML, and financial institutions are often in the middle of this activity. Like any other illicit business, human traffickers use money service businesses, cash, credit cards, and bank accounts to launder money. Financial institutions serve as the vehicles to facilitate this flow of funds, which means that they are at the front lines of detection and prevention. The current anti-money laundering (AML) environment that relies on simple transaction-monitoring rules to detect human trafficking simply does not have the capacity to consider, weigh, and examine the necessary number of inputs that are associated with human trafficking. In practice, this results in a countless number of false positives and missed opportunities. This creates a challenge for financial institutions as they face negative backlash from the public if they are caught aiding, even unintentionally, human trafficking activities.
Researchers and data scientists generally agree that we are in the “age of implementation” for AI/ML. AI/ML researchers are exploring opportunities to deploy the relatively new technology as a tool to combat human trafficking. For example, a team of computational researchers at Lehigh University's P.C. Rossin College of Engineering and Applied Science have joined forces with policy experts, law enforcement officials, activists, and survivors to leverage the same AI/ML components used to track our online preferences (data mining, text mining, and graph mining), to watch for the illicit behaviours associated with human trafficking. Traffickers use the internet and social media platforms to recruit victims and advertise to potential customers, all of which generates electronic data ripe for analysis. A similar technology called Traffic Jam, created by Carnegie Mellon University graduate Emily Kennedy and a team of machine learning experts, leverages AI/ML to systematically scan images of missing individuals and cross-references those images with online advertisements for sex.
These are just a few of the new and emerging applications that researchers, policy experts, and law enforcement agencies have been introducing into this space. In the long term, analysts project that AI/ML will save the banking industry more than $1 trillion by 2030 and estimate that the banks that have incorporated AI/ML into their AML programs have experienced a 20%-30% decrease in the rate of false positives and a threefold improvement in their alarm-to-suspicious conversion rate.
In summary, AI/ML can help financial institutions in the fight against human trafficking and can be used to illuminate patterns that human investigators and manually programmed computers cannot or have not been able to detect. AI/ML can convert the simple and overly exclusive models that financial services employ today to a finely tuned, data-driven mechanism capable of learning from terabytes of data to identify the nuanced trends attributed to different forms of human trafficking.
Additional contributors: Balki Aydin and Joseph Tawney.