According to the United Nations Office on Drugs and Crime, financial services companies globally are penalized to the extent of USD 35 Billion annually for non-compliance. This despite ~USD 100 Billion being spent every year on financial crime compliance
Intelligent segmentation of people and behaviors holds the key to positive Anti-money Laundering (AML) and Know Your Customer (KYC) outcomes – which goes beyond mere customer knowledge or information
Artificial Intelligence (AI) and Machine Learning (ML) make this possible with minimal impact to existing systems – at both speed and scale
USD 4.38 Trillion or 5 percent of the global GDP. This, according to the United Nations Office on Drugs and Crime, is the amount of money that is laundered annually. This office further estimates that while globally the financial system spends close to USD 100 Billion a year on compliance (and USD 15-20 Billion on enterprise software and analytics), companies are still penalized to the extent of USD 35 Billion for non-compliance.
In an ever-changing landscape of financial fraud, money laundering has constantly evolved to remain under the radar of detection. Traditional rule-based and statistical approaches have proved inadequate, slow, laborious and expensive. The crucial need is for newer approaches to detection, and newer tools, technologies and practices.
The key to attacking the Anti-money Laundering (AML) menace lies in intelligent segmentation across its lifecycle. This goes beyond mere customer knowledge or information – and certainly beyond the diligence level. This is where Artificial Intelligence (AI) and Machine Learning (ML) play a vital role. When layered over trusted data from different sources, they can effectively streamline and increase the visibility on risks, based on behaviors through the Know Your Customer (KYC) / AML lifecycle.
At the onboarding level, AI and ML add immense value to the KYC process. For example, image recognition features can verify if the documents match by scanning various sources to evaluate a customer’s KYC risk score. Customer data that is extracted can be seamlessly transferred into customer onboarding systems. A self-learning solution can create dynamic questionnaires that adapt to customers’ responses and map them to intelligence gathered from other sources. Additionally, based on the mapping, chatbots can be deployed to seek additional KYC documents. Such robust automation enhances process effectiveness and achieves significant cost savings.
At the operational level, the combination of Robotic Process Automation (RPA), Natural Language Processing (NLP) and cognitive computing is a potent positive factor. Sophisticated clustering techniques, for example, can group individuals by multiple parameters for highly reliable KYC risk assessment. This information can be leveraged to screen diverse money laundering risk signals, match true identities, identify complex money laundering patterns and provide early alerts.
AI and ML solutions can also conduct real-time analysis on transactions, their history and on multiple unstructured data sources. This will unearth opaque links between entities, and offer a host of cryptic money laundering clues (such as SWIFT messages, and anomalies in invoices and addresses) to more effectively identify Ultimate Beneficial Owners (UBOs) and Politically Exposed Persons (PEPs).
Constructing behavior archetypes and re-classifying risk ratings thus become possible with speed and accuracy. Embedding predictive behavior analytics can enable the capture of a larger number of risk signals, while self-learning models can lead to real-time transaction monitoring.
At the reporting level, AI and ML solutions demonstrate continuous learning to constantly identify and map dynamic alert models based on behaviors. Be it rogue employees, insider trading, benchmark rigging, or other forms of market manipulation, they can analyze entire trading portfolios in real-time to compare transaction behaviors and patterns. This will reveal the trader’s intent – present and future – to commit fraud and market abuse. By leveraging NLP and cognitive capabilities, banks and Financial Institutions (FIs) can also automatically seek, analyze and implement regulatory changes and revisions.
RPA and ML features have the ability to proactively map amendments with products, services and processes to generate relevant workflows. For example, a leading Australian bank piloted an NLP-based AI solution that converted 1.5 million paragraphs of regulatory content with an incredible accuracy level of 95 percent into action points for compliance – in just two weeks.
Additionally, AI and ML enable banks and FIs to create dynamic dashboards of scenario analysis for tax forecasting and reporting – and even propose the right tax strategies. They can bring greater transparency and accountability in data aggregation, analysis and reconciliation, thereby ensuring better compliance.
Intelligent segmentation of people and behaviors holds the key to positive AML and KYC outcomes. It achieves better detection of both intent and action. It also enables more effective benchmarks in rules and thresholds. AI and ML make this possible with minimal impact to existing systems – at both speed and scale. The bonus is that they get smarter with time. Above all, the transparency offered by these technologies make them totally compatible with compliance requirements to significantly slash penalty costs. Little wonder AI and ML solutions have got the ‘thumbs up’ from both regulators, banks and FIs.
Join the conversation
29 October 2019
Finance and Accounting
05 December 2023
Insurance; WNS Triange
30 March 2023