The current financial crime compliance management efforts are based on a combination of automated but static analysis of a pre-determined set of risk factors, together with human judgement. Legacy systems are updated with new algorithms and manually inputted information, generating matrixes for risk interpretation and action, but these very rarely offer a real time overview of risks.

Traditional tools do not allow data to be analyzed at a large scale, limiting the potential for correlations and analysis to generate a more fine-grained picture of the risks. In addition, the quality of the data obtained by legacy systems varies and may not offer the accuracy and detail required to comply with anti-money laundering (AML) and counter terrorist financing measures (CFT) standards.

How Innovative Technologies Can Help –

As discussed above, one of the main challenges hindering the effective implementation of AML/ CFT measures is poor understanding of ML/TF threats and risks. Decision-making, based on inadequate risk assessments is sometimes inaccurate and irrelevant, relying heavily on human input and defensive box-ticking approaches to risk, rather than applying a genuinely risk-based approach. This is where new technologies can provide the most added value and have the potential to AML/CFT measures faster, cheaper and more effective.

Identity verification – Digital identity solutions can enable non-face-to-face customer identification/verification and updating of information. They can also improve authentication of customers for more secure account access and strengthen identification and authentication when onboarding and transactions are conducted in-person, promoting financial inclusion, and combating money laundering, fraud, terrorist financing and other illicit financing activities. They can minimise weaknesses in inconsistencies related to human control measures, improve customer experience, improve customer experience, generate cost savings, and facilitate transaction monitoring.

Traditional ID requirements may be the most obvious instrument to identify customers but should not be the only tool used for this other similar instrument may be more beneficial to the CDD process than forcing in-person production of physical ID documents, notwithstanding the role and review of human analysts and experts which remains key to prevent bias and other unintended consequences of over-technology reliance.  

Screening - The enhanced use of technologies, for client screening and matching, holds great potential to improve the compliance processes, as reliance on out of date and regionally irrelevant sanctions’, PEP and other lists are acknowledged as an area in need of improvements. Such tools allow differentiation of similar names and other elements of identification, overcome language differences, identify cross-references with adverse media information and different databases.

Natural language processing and fuzzy matching tools also allow for a more efficient reduction of false positives and negatives (e.g. in sanction screening processes) but chiefly overcomes problems of data quality, as the programmes become better at linking elements of information, for example, connecting search engine results with PEP lists, identifying fraud attempts, monitoring sanctions lists, etc.

Customer due diligence - Onboarding tools that allow for quick Customer Due Diligence (CDD) and client traits analysis (such as geolocation, credit checks, anti-fraud and others) would also enrich the CDD and monitoring process and lead to a more accurate understanding of the nature of the business relationship, as well as its impact to the institutions. Enhanced ongoing monitoring over an extended period, and behavioural analytics, can give a more robust basis for customer risk profiling and improve the effectiveness of enhanced due diligence related to the lack of trustworthiness of customer identification and verification, potentially allowing to extend the functions of the accounts.

Better risk assessments, CDD procedures and adequate monitoring tools could become an important part of more inclusive and safe financial systems that do not discriminate based on, social or regional context. Risk analysis generated by CDD is too rule-based rather than behavioural or contextual leading to the financial exclusion of unprivileged individuals or groups, who struggle to comply with the requirements.

Transaction monitoring - Artificial intelligence (AI) and machine learning (ML) technologies applied to big data can strengthen ongoing monitoring and reporting of suspicious transactions. Can automatically monitor, process, and analyse suspicious transactions and other illicit activity, distinguishing it from normal activity in real time, whilst reducing the need for initial, front-line human review. AI and machine learning tools or solutions can also generate more accurate and complete assessments of ongoing customer due diligence and customer risk, which can be updated to account for new and emerging threats in real time.

Machine learning are useful for detecting anomalies and outliers identifying and eliminating duplicate information to improve data quality and analysis. For example, Deep Learning (DL) is an advanced type of machine learning in which artificial neural networks (algorithms inspired by the human brain) with numerous (deep) layers learn from large amounts of data in highly autonomous ways. DL algorithms perform a task repeatedly, each time tweaking it a little to improve the outcome, enabling machines to solve complex problems without human intervention.

Global internet research (web scrapping) - Scanning of public data for analysis, construction of indicators and/or formation of databases to extract information related to suspicious entities involved with ML/TF is critical process. This is where, machine learning can be used to read the news and extract from them evidence of legal entities involved in ML/TF activities.

Regulatory reporting - Application Programming Interface (APIs) and Distributed Ledger Technology (DLT), data standardisation, and machine-readable regulations can help regulated entities report more efficiently to supervisors and other competent authorities. The technologies also allow alerts, report follow-ups, and other communications from supervisors, law enforcement, or other authorities to regulated entities and their customers, as well as communications among regulated entities, and between them and their customers. The application of more advanced analytics by regulators can also strengthen examination and supervision, including by potentially providing more accurate and immediate feedback.

Virtual assets - Unlike transactions through conventional intermediaries such as banks, transactions of virtual assets (VA) based on DLT are often conducted without the use or involvement of intermediaries and other obliged entities, and they face obstacles to achieve regulatory objectives, especially those related to AML/CFT, due to the difficulties in tracing and monitoring transactions that may derive from its unique nature. As virtual assets become more widespread, the risk mitigation by intermediaries may become challenging over the medium to long term.

Supervisor assessment - An API-based AML data architecture and AI-driven analytics tool can be leveraged for a centralized platform to generate standardized, automated requests to supervised entities with raw data received through push or pull submission stored in a data lake. An API which can be helpful to establish secure, direct line of machine-to-machine data transmission feeding the data into a processing engine instantly running validations tests verifying quality, content and structure of reports and funnelling processed data into the data lake creating a consolidated, single, and access-controlled data architecture. AI-driven analytics that detects suspicious transactions using predictive analysis and ML techniques (clustering, neural networks, logistic regression, random forests) and recommend AML.

Conclusion 

The use of new technologies in the identification, assessment and management of ML and TF risks allows risk analysis to be more dynamic, provide network analysis, and operate at customer, institutional, jurisdictional, and cross-border levels. Because available technologies can help:

1. Process large volume - A greater capacity to collect and process data, as well as share it among stakeholders, could offer significant advantages in this area, as it would promote a more dynamic risk-based approach.
2. Improve degree of confidence - Even when, the conclusions reached using such tools are the same as those resulting from traditional risk analysis, this confirmation can reassure actors of the completeness and accuracy or their assessments. In this way, machine learning can increase the degree of confidence when applying risk-based measures – and allow them to justify the use of such measures more comfortably to their supervisors.
3. Remove human error – Both digital ID and AML compliance transaction monitoring tools - can facilitate more accurate and up-to-date risk assessments at an optimised cost and provide greater confidence in the conclusions of that risk assessment, enabling greater use of simplified due diligence where appropriate.

And optimal use of these tools factoring regulatory and policy environment that frames adequate data pooling and sharing, or collaborative analytics, which can be appropriate access by supervisors and law enforcement can be significant in identifying emerging threats & stopping ahead of time.