How AI will shape the future of AML protocols

Black piggy bank
Image credit: Depositphotos

Global occurrences of money laundering are on the rise. With every new regulation put in place to combat financial crimes, criminals develop ever more sophisticated workarounds, forcing financial institutions (FIs) to come up with new anti-money laundering (AML) solutions year after year.

Artificial intelligence (AI) is emerging as one of the most highly effective tools in helping banks prevent and combat financial crime. This is great news since under 0.2 percent of laundered money is actually reclaimed.

According to Thomson Reuters, in the U.S. alone, FIs are currently spending about $8 billion a year on AML compliance. AML breaches are not only costly, but they also have far-reaching consequences for FIs and economies as a whole.

The problem with current AML technologies

Many current AML solutions struggle to keep up with increasingly onerous regulations. Most AML technologies are rule-based. And as more rules are added into the mix to counter fraudulent activity,  more cases get flagged for further inspection.

AML monitoring technologies generate over 95 percent false positives because assumptions based on rules are far from reality and data-related glitches can easily disrupt the AML program data.

AI can help by detecting hidden patterns easily missed by its human counterparts and can create accurate segments leading to better rules/thresholds for the AML program. This dramatically lowers the rate of false positives and saves time and effort previously expended reviewing cases.

Learning to trust AI

Where is the trend toward using AI-driven technologies to detect fraud headed in the next ten years? First, let’s take a look at where we are today.

Fear and mistrust of rapidly evolving AI technologies are among the biggest obstacles to AI becoming the go-to solution for preventing money laundering. But even if banks are initially apprehensive about trusting these newer technologies, huge profit losses are the trend that will eventually drive the change.

In a few short years, those who can’t fathom Alexa’s intrusive presence in their homes today will learn to accept her because they won’t be able to function without it or some similar technology.

That same adjustment—from mistrust to can’t-do-without—will be made on a larger scale by financial institutions as they necessarily begin to rely more on AI to prevent fraud and other criminal activities.

The future of fraud detection: Context is everything

The AML landscape will look different in ten years, but how?

Innovations in AI are shifting in focus from improving human intelligence to being more imitative of humans’ decision-making capabilities.

When it comes to detecting fraud, context is important. People and entities must be examined together with all of their connections and associations—and not in a vacuum.

AI-based AML technologies can monitor transactions, screen for sanctions and do know-your-customer (KYC) checks around the clock. Machines have unending resources they can use to make sense of disparate datasets and information silos. They then use this information to create comprehensive overviews of people, entities and all of their associates.

Humans get tired but AI never does. It can place entities fully into context, catching things humans often can’t. What is glaringly obvious to AI might be easily missed by its more fallible human counterparts.

AI is “Extra Human”

The evolution of capabilities that detect patterns and trends will continue to grow AI technology. Not into a superhuman force, but into something even better.

AI’s ability to tirelessly do tasks that would take thousands of humans to carry out effectively, coupled with its native ability to identify trends, makes it a formidable ally in the fight against money laundering.

The addition of what we’ll refer to here as “extra-human” capabilities—an expansion on typical human intuition and decision-making abilities—is what will take AI-based fraud detection to the next level.

Extra human just means hypervigilant. Humans on a larger, more sensitive scale, exercising diligence that would wear out actual people.

2029, An AI Odyssey

Financial criminals hide behind the good reputations of other people and institutions. Extra human intuition and decision-making skills will allow AI to detect connections, and investigate relationships between individuals, institutions and their accounts and locations.

For example, using insurance policies to launder money is commonplace. The criminals that buy the policies are shielded from detection by the insurance agent who is known by the providing financial institution as a reputable agent.

In one insurance case, clients in several different countries went through an intermediary to buy insurance using identification that was never verified by the financial institution.

Eventually, the policy would be closed out and a payment made to a third party.

In a situation like this, AI could identify minute deviations from the insurance agent’s ordinary behavior that would identify the transactions as fraud.

Imagine augmenting AI’s powerful ability to sense trends and put people and transactions into context with natural language processing (NLP), voice recognition and face recognition technologies.

Criminals and terrorists may well be able to find workarounds for all of these applications individually, but gathered together into one force?

Well, bad actors are persistent characters, so it will likely be a case of, “the difficult we do immediately. The impossible takes a little longer.”

With the amount of money laundered at 2 to 5 percent of global GDP, a little longer is a pretty good step in the right direction.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.