Driving AI in Fraud and AML Compliance for Financial Services – with Nick Lewis of Standard Chartered Bank

In an increasingly digitized world, it’s becoming increasingly challenging for banking leaders and law enforcement to combat financial crimes, such as fraud and money laundering. As technology advances, criminals continue to keep pace and employ increasingly sophisticated approaches.

Data privacy laws, while essential and beneficial for law-abiding customers, limit financial organizations’ ability to carry out basic business operations. Furthermore, banks face pressure to maintain regulatory compliance while striving to maintain operational efficiency. 

According to the International Monetary Fund, money laundering scandals erode public trust and have even caused bank collapses. While it can be challenging to precisely quantify the impact on public trust, the amount of money laundered can serve as a good proxy. 

For example, failures in TD Bank’s anti-money laundering policies facilitated three money laundering networks that transferred $670 million through TD Bank accounts. Despite the efforts of the Financial Action Task Force to outline a clear framework for practical anti-money laundering measures, banks still face significant obstacles when attempting to catch criminals across borders.

Emerj Senior Editor Matthew DeMello sat down with Nick Lewis from Standard Chartered Bank on the ‘AI in Business’ podcast to continue their conversation about the continued transformation of anti-financial crime efforts.

The following article will focus on three key takeaways from the conversation:

• Leveraging deterministic AI systems for financial crime detection: Acknowledging the necessity of human input in addressing the limitations of simple, rules-based AI systems for distinguishing legitimate from criminal behavior while detecting anomalies in financial transactions. 
• Balancing AI automation with human oversight in banking: Emphasizing the need for human judgment in financial crime investigations to overcome limitations related to confidentiality concerns.
• Addressing global financial crime through improved information sharing: Highlighting the need for faster, cross-border data sharing among banks and law enforcement to combat money laundering networks.

Listen to the full episode below:

Guest: Nick Lewis, Managing Director, High Risk Client Unit, Standard Chartered Bank

Expertise: Leadership, Risk Assessment, Threat Mitigation

Brief Recognition: Nick worked in UK law enforcement for over 30 years and led several significant operations. He served as Counsellor, Transnational Organised Crime, at the British Embassy, Washington, DC, between 2008 and 2013. In March 2013, the US government awarded him the National Intelligence Medallion, and in 2014, Her Majesty the Queen awarded him an OBE for “services to international law and order.”

Leveraging Deterministic AI Systems for Financial Crime Detection

Lewis begins by emphasizing that crime is not an abstract concept. He describes crime as a series of activities that are not only highly traceable and mappable but also predictable. He points out that many criminals are likely to commit their crimes using a repeat pattern of events based on the idea that if something worked for them yesterday, they are likely to do the same thing next time. 

He explains that the financial sector traditionally responded to that by developing red flags that pick up predictable patterns of behavior and then tune them into rules-based systems. Individual rules are established and used as a basis for generating red flags. Next, organizations create alerts based on those red flags and investigate those alerts to determine if they’re suspicious. 

Lewis cautions that red flag behavior doesn’t necessarily indicate criminal behavior. He sees the way to begin distinguishing between legitimate and questionable behaviors is to use AI to analyze human behavior characteristics and map typical customer behavior patterns, thereby determining whether a behavior is anomalous within a range of norms. 

An example of anomalous behavior that AI can help identify might be if an individual who has a single employer suddenly receives deposits into their account from multiple sources, marked as salary. Lewis explains that combining those alerts with publicly available information from sources such as an individual’s LinkedIn profile can help companies understand the context of those alerts. Context is everything in financial crime, according to Lewis.

Lewis succinctly surmises that “a transaction is a snapshot of a second of one dimension of one element of the client’s behavior on that day.”

Balancing AI Automation with Human Oversight in Banking

When asked about what is an effective way for banks to balance investment between developing AI systems to detect unusual banking transactions and maintaining and investing in human oversight to decide when to involve law enforcement, Lewis explains that he sees it as a three-step process:

• Step 1: Utilize legacy rules-based, analog-like methodologies to detect deviations from known rules and accepted behaviors, providing a foundation for identifying anomalies.
• Step 2: Apply machine learning to identify alerts that are unlikely to indicate financial crime
• Step 3: Utilize AI to predict risks by integrating diverse data points, providing context to interpret the transaction and the client’s behavior.

Lewis goes on to explain that the banking world is mindful about AI for two main reasons. First, AI tools typically perform best in open-source environments; however, banks can’t expose sensitive client data to such environments due to confidentiality requirements, and they can’t import vast amounts of external data into their systems. 

These limitations necessitate a cautious approach to AI deployments, one that not only safeguards client data but also filters out irrelevant Internet noise. Second, while AI can automate manual tasks and reduce staff requirements, it cannot still apply judgment in the same way a human can, particularly in complex financial crime cases. 

According to Lewis, the issue with AI is that it dislikes not being able to provide an answer, so it will gather all the information it has and offer what it believes is the best possible answer, but it won’t acknowledge uncertainty. The ‘false confidence’ from AI interactions is the reason Lewis cautions that organizations need to be cautious, ensuring they’re not using AI to make decisions that would differ from the judgment humans would make. As a result, Lewis predicts that there will always be a need for human investigators, particularly in highly nuanced, complex cases.

Addressing Global Financial Crime through Improved Information Sharing

Lewis disagrees with the default assumption that law enforcement has an advantage in investigating financial crimes. He explains that criminals actually have a massive advantage over law enforcement because they don’t respect borders and therefore are not impeded by national or jurisdictional barriers, unlike banks and law enforcement. 

Even large global banks face restrictions when sharing data internally across countries. As a result, it’s nearly impossible for banks to report the complete picture of international money laundering networks to any single law enforcement agency. 

“For big global institutions like ours, we are also hampered and bound internally by our ability to share information and share data across borders, even within our own institutions, let alone share information with another institution,” he tells the Emerj podcast audience.  

These difficulties persist despite the mandatory obligations banks have to help law enforcement, particularly regarding money laundering networks: 

“I can’t tell one single law enforcement agency. I can’t tell one single law enforcement agency. I have to deconstruct the whole network and tell the FIUs or law enforcement in each of those individual countries. I have to tell them the bit that applies to them. In some cases, I’m even prevented from telling them that there is a dimension in another country.”

– Nick Lewis, Managing Director in the High Risk Client Unit at Standard Chartered Bank

Lewis explains that to combat global financial crime effectively, leaders in both industry and law enforcement need to advocate for systems that make it easier and faster to share information across borders, thereby improving the marked disadvantage that financial institutions face compared to the criminal enterprises they defend themselves against.