We all know that an excess of false positives results in a huge waste of time for AML investigators, which translates to significant and unnecessary cost. It also impedes a bank’s ability to target legitimate money laundering events.
But is the attention we pay to false positives causing us to overlook the hidden dangers of false negative alerts?
While more attention tends to be given to false positive alarms, false negatives can be much more dangerous. Simply defined, a false negative indicates there is no crime, when in fact there is one. Such alerts are overlooked because, as far as investigators are concerned, they don’t exist.
It’s estimated that anti-money laundering (AML) programs have less than 0.1 percent impact on criminal financial crime. This makes the impact of false negative alerts even more concerning.
A primary worry to financial institutions should be the effect false negatives have on their own financial well-being. Federal regulations call for stiff penalties when criminal activity goes undetected. Not catching false negatives greatly increases the incidents of these hefty fines. And when banks are penalized for not adhering to FinCrime detection laws, they risk damaging their credibility with both the industry and consumers.
As criminals become more technologically advanced, they learn how to avoid most red flags of conventional detection systems. And as crime, transactions and data volumes continue to increase, false negatives play an even bigger role in making it harder for banks to find and report real crime…and authorities are not particularly apt to accept excuses.
Many experts agree that some of the most-common causes of false negative alerts include system shortcomings, lack of data and user error. But often overlooked is that when we tighten rule thresholds to reduce the number of false positives, we inadvertently increase false negative alerts. So, in trying to solve one problem, we create another.
As with the misrepresentation of false positives, traditional rule-based detection systems are notorious for getting false negatives wrong. They apply static rules in a dynamic-risk world, where money launderers utilize ever-changing methods for inserting and layering the origins of unlawful funds. Traditional transaction monitoring systems with 1-3% SAR ratios cause investigators untold sleepless nights worrying about risk that’s going undetected.
Unfortunately, a large segment of the banking sector relies upon outdated AML methodology to detect criminal activity. What must be done to reduce false negatives is the same thing needed to reduce false positives…tackle the problem holistically with entity-based, supervised and unsupervised machine learning. By doing so, banks can successfully identify both known and unknown suspicious behavior. Accounts that once appeared to be unrelated are correctly flagged as interconnected, greatly increasing relevant alarms, while reducing both false positive and false negative alerts.
As mentioned earlier, less than one percent of all laundered money is successfully intercepted by authorities each year. Existing transaction monitoring systems are way too antiquated to do the job. AML teams must embrace a machine learning approach that is precise and can’t be fooled in a predictable way by criminals. That is how the industry will reduce false negatives and create risk-based compliance programs that finally give FinCrime a run for its money.
We all know that an excess of false positives results in a huge waste of time for AML investigators, which translates to significant and unnecessary cost. It also impedes a bank’s ability to target legitimate money laundering events.
But is the attention we pay to false positives causing us to overlook the hidden dangers of false negative alerts?
While more attention tends to be given to false positive alarms, false negatives can be much more dangerous. Simply defined, a false negative indicates there is no crime, when in fact there is one. Such alerts are overlooked because, as far as investigators are concerned, they don’t exist.
It’s estimated that anti-money laundering (AML) programs have less than 0.1 percent impact on criminal financial crime. This makes the impact of false negative alerts even more concerning.
A primary worry to financial institutions should be the effect false negatives have on their own financial well-being. Federal regulations call for stiff penalties when criminal activity goes undetected. Not catching false negatives greatly increases the incidents of these hefty fines. And when banks are penalized for not adhering to FinCrime detection laws, they risk damaging their credibility with both the industry and consumers.
As criminals become more technologically advanced, they learn how to avoid most red flags of conventional detection systems. And as crime, transactions and data volumes continue to increase, false negatives play an even bigger role in making it harder for banks to find and report real crime…and authorities are not particularly apt to accept excuses.
Many experts agree that some of the most-common causes of false negative alerts include system shortcomings, lack of data and user error. But often overlooked is that when we tighten rule thresholds to reduce the number of false positives, we inadvertently increase false negative alerts. So, in trying to solve one problem, we create another.
As with the misrepresentation of false positives, traditional rule-based detection systems are notorious for getting false negatives wrong. They apply static rules in a dynamic-risk world, where money launderers utilize ever-changing methods for inserting and layering the origins of unlawful funds. Traditional transaction monitoring systems with 1-3% SAR ratios cause investigators untold sleepless nights worrying about risk that’s going undetected.
Unfortunately, a large segment of the banking sector relies upon outdated AML methodology to detect criminal activity. What must be done to reduce false negatives is the same thing needed to reduce false positives…tackle the problem holistically with entity-based, supervised and unsupervised machine learning. By doing so, banks can successfully identify both known and unknown suspicious behavior. Accounts that once appeared to be unrelated are correctly flagged as interconnected, greatly increasing relevant alarms, while reducing both false positive and false negative alerts.
As mentioned earlier, less than one percent of all laundered money is successfully intercepted by authorities each year. Existing transaction monitoring systems are way too antiquated to do the job. AML teams must embrace a machine learning approach that is precise and can’t be fooled in a predictable way by criminals. That is how the industry will reduce false negatives and create risk-based compliance programs that finally give FinCrime a run for its money.
