Risk and Reward: AI Enables Active Financial Crime Risk Assessment

Adverse media is openly available, continually updated, and provides a wealth of information, but it remains underused in financial crime risk assessments, writes Daniel Banes.

Artificial Intelligence (AI) adoption by banks to power their ongoing mission to protect the financial system against financial crime has advanced from “scrutinised” to “accepted” to “expected” in just a few years.

The Monetary Authority of Singapore (MAS) and Hong Kong Monetary Authority (HKMA) have each released principles for the use of AI in the financial sector[1] and have continued to engage the industry through forums, circulars, newsletters, speeches, and guidance papers on the application of AI for AML/CFT[2].

The message is clear: the regulators are seeking ways to inspire the financial industry to fully embrace AI and data analytics for more efficient and effective programs.

By any measure, there is room for improvement in current financial crime risk detection and mitigation effectiveness. Although large investments and enhancements have been made over the last several years, there has been a marginal impact on money laundered, evidenced by the amount of criminal proceeds forfeited compared to the estimated funds laundered globally (2 to 5 percent of global GDP, estimated by the United Nations Office on Drugs and Crime[3]).

This should come as no surprise: Banks will continue to play catch-up until they are able to effectively leverage available data to fight financial crime – a steer that regulators continue to urge.

Blind spots in plain sight

One of the key financial crime risk data points is openly available and continually updated yet commonly underused: adverse media. To make a dent in the trillions of dollars laundered through the financial system each year, we need to better harness the full global effort to expose criminal behavior; that includes the investigations and reporting by journalists, governments, and publications by local media outlets.

In the digital age, the low-hanging fruits of their labour are available on the open web in real-time. While banks have incorporated some adverse media analysis in their AML/CFT processes, current controls tend to lack review frequency and coverage; two critical elements needed to take full advantage of this wealth of information. With proven AI-powered tools now available, realising the effectiveness benefits from enhanced frequency and coverage will soon be “expected” for compliance programmes.

Frequency. Bank customers’ lives are dynamic and adverse media is correspondingly updated, yet banks continue to rely on static data retrieved at customer onboarding or periodic review to make financial crime risk assessments. This is a fundamental flaw –  a risk assessment made on today’s information is already out of date once a new event is reported tomorrow.

Under the current process, that reported event—instantly available for assessment—awaits discovery until the point of the next customer review, whether that be in six months when a review is triggered by another financial crime control, in a year when the scheduled periodic review occurs, or even longer for perceived medium and low-risk customers.

So, even if a material financial crime is reported the day after a customer’s onboarding, that knowable risk remains unnoticed, allowing a criminal unfettered access to the financial system for an extended period of time.

Coverage. From the front-page headline stories published by global media outlets to the back-page articles in little-known local newspapers, adverse media is reported in large and small markets, in various languages, every day. While it is expected that relationship managers would notice global headline news related to their customers, some banks look to a single data or watchlist provider to uncover the rest.

However, there is both an ever-increasing time gap between the point at which an adverse event is published in the digital domain and the time it takes for that event to appear on a watchlist, and an information gap between the data available on a watchlist and the information available from across the digital domain. Watchlist data, while useful as one of several sources, only covers a fraction of the total sum of data available, putting a ceiling on programme effectiveness if used alone.

Unlike relying on stale or limited information, a perpetual customer risk profile—that incorporates ongoing adverse media monitoring with suitable coverage—enables proactive risk management. Many banks have not yet started along this path and muscle memory would dictate inflated staffing requirements to implement such a process with an inundated front line upon process deployment. However, programme overhaul is not required when implementing a combination of purpose-built technology, a calculated process, and human expertise.

Harness information at scale

Technology allows the industry to harness information at scale and unite the broader global effort to thwart financial crime, thereby empowering banks with a more effective and current financial crime compliance programme without giving up coveted efficiency gains. AI’s ability to automate live content retrieval and use natural language processing to assess content across languages, get rid of false positives and duplicates, and categorise relevant information by risk type allows for efficient, effective, and continuously calibrated risk analysis for customers.

Pairing this technology with financial crime risk professionals that understand industry standards and materiality relieves excessive staffing review needs and concentrates front line risk owners’ effort to assessing truly risky content. Embedding these key components in a quality-assured process that provides actionable risk and operation insight helps to sustain the effectiveness uplift and actively procure continued efficiency gains.

A perfect storm of events has created a persuasive atmosphere for banks to decisively improve AML/CFT programme effectiveness and action the lessons learned from continued cases that could have been detected simply by leveraging the digital domain: proven technology has emerged, a global pandemic has strained operations and fast-tracked technology implementation[4], and regulators are encouraging AI adoption.

Both the HKMA and MAS continue to report examples of successful AI application in AML/CFT programs, showing both efficiency and effectiveness gains. Likewise, regulators around the world have touted their own adoption of AI and advanced technology to root out bad actors, encouraging financial institutions to do the same, and paving the way for adverse media monitoring to become a standard AML/CFT programme upgrade.

Daniel Banes is a Managing Director and the Asia Pacific Regional Leader at Exiger, based in Singapore. He has extensive experience using data analytics and advanced technology to investigate financial crime and improve programmes around the globe.

[1] https://www.mas.gov.sg/publications/monographs-or-information-paper/2018/FEAT;
[2] For example, https://www.mas.gov.sg/news/speeches/2019/combatting-financial-crime-through-new-technologies-built-on-strong-fundamentals; https://www.hkma.gov.hk/eng/news-and-media/press-releases/2019/11/20191122-4/.
[3] https://www.unodc.org/unodc/en/money-laundering/globalization.html
[4] https://www.hkma.gov.hk/media/eng/doc/key-information/guidelines-and-circular/2020/20200730e1.pdf

To Top