With trade and economic sanctions becoming an ever more popular tool of foreign policy in today’s uncertain geopolitical climate, AML, screening and anti-fraud obligations are increasing in scope and complexity. At the same time, the growth in international cross-border trade to around $16 trillion per annum creates an environment that’s ripe for abuse for those wanting to launder money or finance terrorism or criminal activities through the guise of legitimate trade.
It’s no surprise regulatory bodies around the globe are stepping up scrutiny around sanction screening and also around trade-based money laundering (TBML). The International Narcotics Control Strategy Report (INCSR) estimates that hundreds of billions of dollars are laundered each year through TBML. Techniques such as falsifying documents, under or over-invoicing, and misrepresenting financial transactions can involve multiple parties and jurisdictions, making such crime extremely difficult to detect.
Regulators, including the UK’s Financial Conduct Authority, the Monetary Authority of Singapore and the Hong Kong Monetary Authority, alongside the International Chamber of Commerce (ICC), issue guidelines and red flag checks to help combat TBML. The red flags define key attributes in trade finance transactions that indicate a high risk and are seen as the global standard for due diligence for which financial institutions must screen and monitor.
The financial cost and reputational risk for any failure to meet national and international financial crime compliance obligations can be high. Fines issued by OFAC (the US Department of the Treasury’s Office of Foreign Assets Control) are significant, with some fines issued to banks over the last decade exceeding several billion dollars. Banks, payment service providers and corporates have been dedicating a significant amount of human and financial resources to cope with the growing complexity of compliance. It’s now time for a different approach.
The value of AI
Artificial Intelligence (AI) is the key to combatting financial crime, ensuring compliance, and lowering operational costs. It’s effective in increasing the accuracy of detection and in reducing compliance costs by drastically reducing false positives. These are particularly prevalent in the area of sanction screening. Banks are still using manual processes to investigate, monitor and rule out 99% of alerts found to be false.
The AI disciplines of Natural Language Processing (NLP) and Machine Learning (ML) are key in increasing the accuracy of financial crime detection, and the use of self-learning capabilities can significantly reduce efforts to manually review false alerts. Augmented Intelligence (helping humans become faster and more efficient at their tasks, rather than replacing them) is also essential.
Harnessing the unique capabilities of AI technology, banks can go beyond the efficiency benefits of simple document management workflow. They can intelligently monitor and rapidly detect fraudulent activities, without the burden of having to employ large numbers of expensive, error-prone and time-consuming human resources to tackle difficult compliance checks manually. By providing this augmented intelligence, trade finance operational and compliance staff can concentrate on real issues requiring investigative skills rather than the mundane manual tasks of reading and checking documents and inputting data into screening tools.
Any AI-based solution must also be explainable and auditable. Banks and corporates need to be confident about the actions being taken and able to provide explanations to auditors and regulators. Using the context surrounding alerts, NLP algorithms can accurately decipher if an alert is true or false and can provide an explanation.
Fighting TBML in practise
Monitoring for indicators of TBML is a highly labour-intensive and costly process. TBML is difficult to detect because of the complexity of trade finance, involving many entities and data sources, and a long-held reliance on paper. The ICC estimated in 2018 that four billion pages of documents are circulating in trade at any one time. These documents are heavy on free-format text and unstructured data and do not lend themselves well to most compliance filters, making it easy for a misrepresented price, quantity of goods or a false customs declaration to go unnoticed.
Unstructured data from the various paper-based trade documents must first be scanned and put into machine-readable text format with the help of optical character recognition (OCR) technology. Once the data is in a format that can be processed and analysed, NLP can be used in combination with knowledge-based techniques to interpret the text, understand the context and derive meaning from it and to extract key trade information automatically. This machine-readable ASCI information can then be further processed by intelligent AI techniques to compare, identify and provide alerts for red-flag indicators and sanctions subjects.
While TBML remains widespread, red flag guidelines and sanction watch lists will continue to evolve and grow. It is only through flexible, intelligent, automated AI-powered solutions providing augmented intelligence that banks can keep up with and adapt to the heightened regulation required to tackle growing international crime.