The global estimates of human trafficking vary because of the secretive and shadowy nature of the crime: the victims are often invisible, they are uncounted for, and they go unreported when trafficked from their home countries. It’s estimated that between 20 million to 40 million people are in circumstances of human trafficking in any given moment. The global revenue, meanwhile, is a staggering $150 billion a year—making this the third largest and fastest-growing criminal activity in the world.
A crucial component of this story is money. It’s estimated that only one percent of criminal proceeds from trafficking are confiscated or disrupted. That represents an extraordinarily high profit margin on a generally low outlay and explains why it is so lucrative for the criminals involved.
Examples of private sector involvement in human trafficking are abundant: traffickers use banks to deposit and launder their earnings; they use planes, buses, and taxi services to transport their victims; they book hotel rooms, AirBnB, and house rentals: integral to sex trafficking; and they are active users of social media platforms to recruit and advertise the services of their victims. Money paid to victims needs to be interdicted by the gangs, extracted, and then laundered. All these activities leave a digital payment footprint which can tell a story of how people have been ensnared, trafficked, and exploited.
This footprint is the Achilles’ heel that the banks and law enforcement use to detect the money laundering activities of the criminals. But, with the rise of new FinTech companies offering many more account and money transfer options, the digital banking landscape that the criminals now operate in has become vastly more complex. The traditional Anti-Money Laundering methods and processes used by the banks and law enforcement are struggling to keep pace.
However, new artificial intelligence enhanced data science solutions are becoming available to supplement the banks existing transaction monitoring systems. AI solutions can provide tactical and surgical analytical answers to the human trafficking problem by identifying red-flag behaviours to assist in predicting where new and undetected human trafficking activity is occurring.
These behavioural analytics solutions are being used to look across networks of accounts, combining contextual risk factors with transaction data to identify crime typologies, and flag suspicious activity more easily to law enforcement.
But there are limitations to what an individual bank can do even with these new capabilities. AML teams can still find themselves operating in silos, only able to utilise the data that the bank itself has gathered. Adding data from other banks and from external sources such as newspaper and media reports would allow a bank to build a far more detailed picture of customer behaviour across entire industry segments. The challenge is how to share this data across the banking sector while remaining compliant with international Data Privacy regulations such as GDPR.
This is the focus of one of the fourteen Pilots being developed by the EU-funded INFINITECH project under the Horizon 2020 programme. The participants in the pilot are from the National University of Ireland, Galway (NUIG), Bank of Ireland, Banking & Payments Federation Ireland (BPFI), IBM Ireland, and Traffik Analysis Hub (TA Hub).
TA Hub is a not-for-profit organisation and is the largest collection of survivor stories of trafficking and exploitation experience. Using IBM’s Watson Discovery which has been specifically trained on human trafficking terms, it applies machine learning capabilities to ingest open sources of data at scale from multiple sources – such as thousands of daily public news feeds.
That’s something financial institutions have never had access to before: pooled data from multiple sources—NGOs, publicly available news via AI and other peer financial institutions. That gives them a view beyond their own internal horizon. Which, in turn, allows them to better focus the microscope.
By providing partners with the ability to pool data assets TA Hub can generate new insights into patterns and hotspots of trafficking incidents. It is a fundamental principle of TA Hub that collaborative data sharing will provide all its partners with greater value than can be achieved in isolation.
In the INFINITECH pilot we develop intelligent technology focused on identifying the financial operations of the ‘business models’ on which human trafficking is based. This is translated into usable intelligence in how trafficking activities manifest themselves in the financial system. Thus, enabling Financial Crime, Risk Compliance, and Fraud Investigation teams to move from a passive warning to a proactive monitoring approach. By applying custom built NLP based AI models, specifically trained to spot direct references (and inferences) to red-flag indicators, it is possible to generate a candidate list, or typology, of indicator combinations that describe the financial markers in a human trafficking incident. Typologies are updated in real-time to reflect the changing patterns and emerging trends in how trafficking groups operate.
These typologies can then be exchanged between banks and be executed against their own data within their own domains, further training and enhancing the typologies. As enhanced versions evolve, they can be shared back to the community of banks. The BPFI in its role as Industry Association will host a centralised library of these red flag typologies and enables these to be shared across the participating banks.
Each of the participating banks get the benefits of the typologies being run against multiple transactional data sets, without the need for actual data to be shared. In this way the project meets its GDPR obligations.
This capability in principle can be perceived as an Anti-money Laundering capability which is the focus of the Bank of Ireland AML team within the pilot. However, the association of large amounts of data and the sharing of typology information is a unique opportunity and can be utilized by any financial institution involved in profile building for support on a range of different financial detection activities, such as identity scoring, credit scoring, and improving customers profiles.
There is a unique opportunity here to unlock the power of data sharing and collaboration on a massive scale by bringing financial information together across sectors and geographies and make real impact at a level that does not currently exist.