Balancing Speed & Safety – The Challenges of AML Transaction Monitoring for PSPs and Fintechs
In a payments environment that is undergoing its steepest shift yet, PSPs and fintechs have their work cut out for them in AML transaction monitoring. . The most widely-used transaction monitoring systems are not fast or agile enough for the nature of today’s business.
On this page, we review commonly-used AML transaction monitoring solutions, comparing traditional tools and methods to newer technologies that promise to make AML transaction monitoring both fast and secure enough to cope with a rapidly evolving payments landscape.
Transaction monitoring is the process of monitoring customers’ transactions (transfers, deposits, withdrawals, and more) in a bid to identify any suspicious behaviors which could indicate the presence of money laundering or other financial crimes.
Any transactions that the system deems suspicious are flagged and passed to a compliance team for further checking. Flagged transactions found to be “true hits” upon further investigation are reported to the relevant law enforcement bodies. Transactions found to be not suspicious, are categorized as “false alerts” and are processed as normal.
The faster financial services institutions can perform transaction monitoring and separate out true hits from false positives, the faster and more seamless a service they can provide their clients.
All financial institutions today – both traditional and the newer fintechs and PSPs – are required by law to have an AML transaction monitoring system in place. As well as protecting companies from financial penalties issued by regulatory bodies, a demonstrated commitment to transaction monitoring reassures customers and clients of the company’s safety and reliability, thereby promoting business growth.
There are a variety of AML transaction monitoring tools and software solutions in use today:
Rules-based – Traditional transaction monitoring software is rules-based, meaning that a compliance team predetermines and programs the rules used to determine which transactions require further investigation. Rules-based systems are known to report large numbers of false positives, which require hours of time and resources to investigate.
AI/ML-based – A new breed of transaction monitoring solutions built around intuitive AI and ML principles do not rely on predetermined rules. These are more accurate at detecting true financial crimes and throw out far fewer (if any) false positives.
By reducing the volume of false positives, AI/ML-based transaction monitoring tools promise both faster and more reliable transaction monitoring.
How Does Transaction Monitoring Work?
Transaction monitoring systems consider the history of each account holder, the customer profile, and any existing red flags. The transactions of customers with a negative history or known red flags will naturally require more scrutiny than customers with a clear record. A good system will be programmed to look for these warning signs and take action accordingly.
Any transaction deemed suspicious by either a rules-based or an intuitive AI-based system will raise a flag indicating that further investigation is needed. Compliance personnel then sift through all the flagged transactions to determine which are true hits and which are false positives. When a true hit is found, a suspicious transaction report (STR)* or Suspicious Activity Report (SARs)** must be sent to any relevant law enforcement bodies to alert them of suspected cases of money laundering or terrorist financing. Any flagged transactions found to be false positives can be processed as usual.
Is There a Legal Requirement to Perform Transaction Monitoring?
Ongoing transaction monitoring is considered an essential element of effective KYC procedures and is a regulatory requirement for the wide range of business sectors that come under anti-money laundering (AML) and counter-terrorist financing (CFT) regulations. Regulatory bodies across the globe are responsible for setting and monitoring financial crime regulations for their domain. Institutions that fail to implement AML programs can be fined or denied a license, resulting in loss of business.
AML Transaction Monitoring & Risk-Based Approach
Regulators around the world are increasingly demanding a risk-based approach to AML transaction monitoring. A risk-based approach, according to FATF, means that “countries, competent authorities, and banks identify, assess, and understand the money laundering and terrorist financing risk to which they are exposed, and take the appropriate mitigation measures in accordance with the level of risk.”
The risk-based approach requires that FIs and PSPs move from post-analysis and investigating transactions flagged as suspicious, to proactive management. This means performing KYC checks and carefully assessing the potential risks involved in working with specific customers and business entities or operating in certain jurisdictions. They will need to carry out institutional risk assessments and build accurate individual customer risk profiles. It will, of course, require a greater level of due diligence when dealing with customers, businesses, and jurisdictions deemed higher risk.
If using a rules-based system, then all the necessary checks and procedures must be programmed into the AML transaction monitoring system. It is, however, very difficult to program rules-based systems effectively. Erring on the side of caution by creating too many rules can result in an excess of false positives and a compliance logistics nightmare. Too few rules and businesses run the risk of allowing financial misdemeanors through the net and being penalized for doing so.
Financial institutions and PSPs need reliable transaction monitoring software that can cope with all the vulnerabilities introduced by today’s digitalized, globalized and cloud-based business environment. A 2021 FATF (The Financial Action Task Force) report – Opportunities And Challenges Of New Technologies For AML/CFT – identified AI and ML as promising new technologies in the fight against financial crimes and described ML as offering a great advantage “through its ability to learn from existing systems, reducing the need for manual input into monitoring, reducing false positives and identifying complex cases, as well as facilitating risk management”.
By cutting through the noise generated by rules-based AML transaction monitoring systems and providing more accurate results with fewer false positives, AI and ML-based transaction monitoring systems represent the game-changer the industry has been waiting for, having the potential to satisfy the needs of both customers and regulators.
Analyzes a staggering array of factors within datasets, learning what is normal and spotting patterns and anomalies without relying on predetermined human rules and suppositions.
Derives conclusions from the data alone (and not from rules programmed by developers), and therefore it can identify patterns and connections – or, “unknown-unknowns” – not detectable by humans.
Higher accuracy and fewer false alerts than rules-based systems means fewer resources are necessary to monitor the output of the system.
Adapts itself to changing realities using only the data itself. Intuitive AI and ML systems continually improve in accuracy as they “learn” more about datasets.
As they don’t need to be programmed (and re-programmed) with rules, intuitive AI systems are faster, easier, and cheaper to onboard.
While AI and ML-based transaction monitoring is not yet a legal requirement, players in the industry are waking up to its potential, and adoption by tier-one banks and other finance-based businesses is growing. Of the 850 financial institutions that participated in an AML technology study run by SAS, KPMG, and the Association of Certified Anti-Money Laundering Specialists (ACAMS), more than half (57%) had already deployed AI/ML into their AML compliance processes, were piloting AI solutions or were planning to implement them in the next 12-18 months.
Regulators too are starting to appreciate AI’s potential in the AML space with FATF dedicating a section of its 2021 report to promoting the potential benefits of AI and ML in the fight against digital financial crimes.
As PSPs and fintechs proliferate, as cross-border global transactions of increasing complexity become more ubiquitous across the world, and as cyber-criminals intensify their efforts to bypass rules and systems designed to protect these transactions, industry players will seek more reliable and effective transaction monitoring tools. At present, only AI and ML hold the potential to provide the layers of security, reliability, speed, and ease that is needed.
Benefits of AI/ML for AML Transaction Monitoring
Business growth – Allows financial businesses to safely grow and expand into new corridors and industries.
Affordable – Keeps compliance costs low.
Manageable – Reduces false positives, keeping daily operations manageable without sacrificing reliability.
Self-learning – Learns from an ever-changing ecosystem, helping businesses spot potential weaknesses before they become liabilities.
Easy quick and low-cost integration – Easy onboarding, minimal learning curve, and maintenance, and overall low costs.
SONAR is a fully scalable cloud-based SaaS AML/CFT solution based on cutting-edge “artificial intuition” ML, delivering 95% investigation-worthy alerts and painting a clear picture of abnormal activity within large data sets to ease the burden of compliance teams.
Intelligent insights into KYC
Provides customer identity insights even in cases where KYC information is lacking e.g. with non-customers.
Developed by world-renowned mathematicians and based on 20 years of academic research into AI.
Can detect anomalies (unknown unknowns) in big data without relying on rules, spotting innovative and unprecedented crimes.
Massively reduces false positives, decreasing costly compliance activities.
Optional rules layer
Supports an optional rules layer in case rules are needed e.g. to meet regulatory demands.
Cloud-native SaaS solution
Adjustable to any cloud environment – public, hybrid, or private. Easy roll-out, even remotely.