In recent months, artificial intelligence (AI) and machine learning (ML) have been making headlines as people start to realise the potential of tools such as ChatGPT, DALL E and others in their personal and professional lives.
This new wave of innovation, driven by Large Language Models (LLMs)/ generative AI, has brought a renewed focus in organisations to leverage AI in a multitude of scenarios. One of the areas AI can have a significant impact is in tackling fraud and money laundering; the results of which can bring valued benefit to customers and enable regulatory compliance.
The true cost of fraud and anti-money laundering (AML) failure
It is a well-established fact that fraud is a significant problem in the world of payments. UK Finance estimated fraud losses across payment cards, remote banking, cheques and APP scams to be £1.3 billion in 2021.
Globally, it is estimated the amount of money laundered in one year is 2-5% of global GDP or between £650 billion – £1.7 trillion.
In addition, fines have increased more than 50%, totalling £4 billion in 2022 due to anti-money laundering infractions, breaching sanctions and failings in their know your customer (KYC) systems, demonstrating the challenges financial institutions have in keeping up with criminals.
Moreover, the victims of financial crimes cross the full range of consumers from vulnerable individuals and small-medium sized businesses to large public and private sector institutions. And the impact of this varies depending on the consumer with some of the most notable being financial, reputational, and psychological.
Against this backdrop, can financial institutions adopt AI and ML to better protect consumers from the harm caused by financial crime through improving their prevention and detection capabilities?
Where does AI have an impact?
AI and ML play a pivotal role in the detection of suspicious activity through pro-active identification of patterns, perpetual monitoring for anomalies, and prescriptive remediation. Above all, AI agility in data processing and profiling significantly increases speed, helping respective teams to quickly react to any threats.
There are five ways AI can help you manage economic crime risk more effectively.
Save, Curate and Share
Save what resonates, curate a library of information, and share content with your network of contacts.
- Improved KYC/customer due diligence (CDD): Financial institutions have already made strides in automating and digitising their KYC/CDD processes, but they can be further enhanced with powerful machine learning algorithms. Techniques such as network analysis can enable financial institutions to better understand and develop deeper knowledge of the risk exposure of not only customers, but also their wider network, allowing for a better targeted approach to risk management.
- Better customer product suitability assessment process: Complete and well-defined risk assessment policies and processes are a vital element in understanding risk exposure and applying controls in a risk-based manner.AI is having an increasing impact on the assessment of risk, with financial institutions developing sophisticated models based on historic customer activity and known indicators of risk to better predict and define the risk posed. The continued development of AI can only improve the definition of the risk faced by financial institutions, enabling a more mature application of risk management.
- Identifying authentic consumers and financial institutions: As per the National Fraud Initiative report produced by the UK Government in 2022, fraud is estimated to account for 40% of all UK crime. The use of AI, such as the application of facial or voice recognition, is a key tool in the evolution of fraud prevention and is continuing to evolve with newer anomaly detection and behavioural algorithms.AI can not only help detect where a fraudster is impersonating a customer, but also where they are imitating a bank or third party with the intention of requesting money as part of a scam. AI could pick up on behaviour akin to examples of fraud and warn the end user of any identified concern.
- Spotting unusual activity and transactions: Alongside traditional rule-based transaction monitoring systems, which rely on static data sources, ML has become an increasingly popular method of both detection and post-alert management within transaction monitoring. Financial institutions are now applying increased focus on the development of robust ML algorithms to address the challenges of AML/ Counter Terrorist Financing (CFT) as the regulatory bodies are encouraging the financial institutions to develop innovative solutions. Within the 2022 report produced by the Bank of England (BoE) and the Financial Conduct Authority (FCA) on the use of ML, the FCA demonstrated their commitment to monitoring the state of ML deployment and the safe and responsible adoption of technology in UK financial services. This supports the continued development in AI/ML, with the use of deep learning and implementation of AI enabling the effective identification of unusual and/or suspicious activity.
- More efficient operational enhancements: A significant proportion of the cost of a financial transaction can be linked to back-end exception management operations. Post-alert application of AI/ML techniques can better support AML/CTF controls, ensuring resources are focused on the areas of greatest need and controls are applied in a risk-based manner. An example of this is the application of ML models in transaction monitoring and analysis, including:
- Rule refinement/ false positive reduction: Identification of repeated false positives, or high spike periods, which could be indicative of an inefficient rule-base. Financial institutions are now adopting advanced techniques to identify common false positives to allow better focus of resources. Where common patterns are identified, this supports a rule-review process, applying the updated rule-base to better manage effective and efficient alerts.
- Post-alert classification: Supervised ML models can be adopted post-alert generation to classify it into priority categories (e.g., high, medium or low). These categories are typically used in case management systems to deploy resources on a risk-based basis while also enabling a targeted approach to alert investigations.
What are the benefits of adopting AI?
Financial institutions can reap the potential benefits that AI offers through increased operational efficiency, optimisation and better risk management. It can also better protect consumers from the harm caused by financial crime through improved prevention and detection.
As perpetrators become increasingly innovative, it is important that financial institutions stay one step ahead. Adopting AI can help tackle much of the fraud that happens today and could take place tomorrow, at the same time reducing costs and time, as well as improving both customer experience and protection.
Please do not hesitate to contact us should you have any questions or want to discuss this topic in more detail.
