Glass Boxes – Under The Hood of AI Transaction Monitoring

July 26, 2022
Next-generation transaction monitoring tools that use AI and ML technologies are getting faster and more accurate, drastically cutting the number of false positives. Yet fear and suspicion around the “black box” nature of AI is still keeping potential adopters away from these technologies. Black box AI refers to a problem in machine learning whereby it is difficult to explain how conclusions were reached – sometimes even by the algorithm designers themselves. This presents a huge hurdle in a stringent regulatory world where influential bodies such as the Monetary Authority of Singapore (MAS) are calling for fairness, ethics, accountability, and traceability (aka. FEAT) when using AI in the financial sector. In today’s climate, transparency is critical, and regulators must be able to determine if results produced by AI transaction monitoring software were influenced by bias or prejudice and that the correct datasets were used. Fintechs must be able to explain their past screenings transparently and comprehensively to any external auditors, but many of the AI-based screening tools in use today make this impossible.  What the industry really needs (and what very few are able to provide) is a “glass box” model with a highly transparent path from data, through the process, to output and in which it is possible to not only explain why certain results occurred but also how the algorithm got there. The Tides are Turning  There is no stopping the march of progress. Fears around black box AI notwithstanding, regulators, analysts, and big players in the financial services industry are promoting AI’s potential to upgrade the effectiveness and efficiency of AML transaction monitoring on multiple levels. In fact, FATF (The Financial Action Task Force) strongly recommended the usage of AI and machine learning for AML and CFT detection, and also recommends the following: Move away from a rules-based approach – With their abundance of false positives and excessive reliance on human input to create underlying rules, these tools are too burdensome and expensive to run, slow to adapt to new realities, and often not sensitive enough to pick up sophisticated or innovative crimes. Adoption of AI transaction monitoring tools by regulators – The high volume of data generated by fintechs and other financial institutions makes the task of the regulatory bodies virtually impossible without newer tools that are more fit for purpose.  Greater use of AI to enhance financial inclusion – AI transaction monitoring is more sensitive and can spot suspicious activities that rules-based systems have been historically unable to intercept. This means that instead of blacklisting entire countries deemed to be too risky, these countries can now be invited back into the world’s financial fold, knowing we have the right tools to provide adequate security, Citizens of these nations will then be able to benefit from conveniences such as online payments and mobile banking which will help break the cycle of poverty. What is Glass Box AI (And is it Too Good to be True)? With regulators actively promoting AI transaction monitoring software, these tools are here

4 Key Criteria For Choosing Your Transaction Monitoring Software

July 24, 2022
Fintechs rely on transaction monitoring software to secure their innovative platforms against increasingly sophisticated financial crimes. They have an obligation both to their customers and to regulators to ensure their financial business is not exploited by criminals. But in the fast-moving and agile world of fintech, they can’t afford to adopt software solutions that slow down the rate of transactions.   With the constant advances in technology and an active and creative enemy, fintech must adopt cutting-edge systems that enable them to reliably and quickly detect, intercept and report any suspicious activities.  The best transaction monitoring software will cope with a rapidly evolving and increasingly complex environment and keep legitimate payments of ordinary customers flowing when adopting the most advanced payment technologies.   Here are 4 key criteria to keep in mind when choosing your AML transaction monitoring software. Regulatory Compliance According to Deloitte, “financial service providers are investing increasingly in intelligent solutions that use modern technologies to reduce the costs of regulatory compliance.” The rapid uptake of AI-based transaction monitoring software by Tier 1  banks is a clear sign of this trend. Regulators are also lauding the benefits of AI AML, with FATF (The Financial Action Task Force) strongly recommending the use of AI and machine learning for AML and CFT detection. This means that a path toward better AML has already been laid.  Adopting AI AML tools is not a legal requirement, but regulators are becoming stricter yearly with compliance requirements. It seems inevitable that fintechs will be under pressure to adopt the highest levels of monitoring in the not-too-distant future. Furthermore, using AI transaction monitoring tools can help prevent AML compliance fines.  Rules-based vs. AI in transaction Monitoring In recent years, FATF has called for a risk-based approach to preventing money laundering.  This means that financial institutions and businesses must “identify, assess, and understand the money laundering and terrorist financing risk to which they are exposed, and take the appropriate mitigation measures in accordance with the level of risk.” In practice, this means that not all customers or jurisdictions are to be treated equally from a risk point of view.  Customers or countries known to be at a higher risk will therefore benefit from more stringent levels of monitoring. The problem lies in defining who is at risk and what stringencies to apply.  Rules-based,  traditional transaction monitoring software means that rules are used to determine which transactions require further investigation. In rule-based systems, the rules must be predetermined and programmed upfront, and it is easy to be either too comprehensive (resulting in too many false positives) or not comprehensive enough (resulting in suspicious activity going undetected). A new breed of transaction monitoring systems using AI and machine learning, on the other hand, can analyze a staggering array of factors within datasets spotting patterns and anomalies without relying on human input. These systems can derive conclusions from the data alone, thereby identifying unknown patterns and connections. Because AI transaction monitoring software learns from the data itself, it operates more effectively

5 Misconceptions About AI-Based AML Programs

July 12, 2022
Written by: Idan Keret Chief Customer Officer ( CCO ) The industry is on a search for new technologies to manage AML compliance, with AI top of mind now that fintech events are taking place again around the world. Hesitation around making the transition from a rules-based to an artificial intelligence-driven solution can be caused by misconceptions about the methodology of an AI-based program.  Here are 5 misconceptions that can stand in the way of making the transition and explanations that can be used when presenting AI as a solution to regulators. Too much dirty data will prevent the transition. Much like a human brain that operates non-linearly, advanced AI solutions are programmed to analyze data from multiple angles and sources. This means that small anomalous or incorrect data present no real deterrent for advanced AI. In fact, advanced AI can bypass dirty data issues such as duplicate entries, misspelled words, or outdated data and enable financial institutions to make the transition to more effective AML systems. My regulator will object. The FATF, the global anti-money laundering watchdog, actually encourages regulators and financial institutions to adopt new and advanced technologies for AML/CTF such as artificial intelligence and machine learning for more efficient and effective use of resources to detect financial crimes. In fact, top 100 banks and top 100 payment fintechs around the world have already replaced rules with machine learning approaches technology.   And of course, AI can improve the quality of SAR submissions. The FATF reports that machine learning is offering the greatest advantage to users “through its ability to learn from existing systems, reducing the need for manual input into monitoring, reducing false positives and identifying complex cases, as well as facilitating risk management.” AI is not explainable.   Every AI method has its methodology. The difference is that AI is math driven, meaning it is precise, logical, and driven by facts that can be verified and validated. In unsupervised AI, algorithms are programmed to identify abnormalities. When applying this methodology to risk factors, the system can detect suspected cases of ML that deviate from norms for the given typology. Therefore, AI cannot only be explained and is far more accurate than rules will ever be. In AI, the computer takes over to intelligently solve problems with no human bias about how things should be. In this way, new and unknown threats can be detected. Going to the cloud is cumbersome and requires extra effort. Cloud-enabled services are typically API-based and fast methodology. Users of cloud-native services gain benefits such as scalability and the ability to receive fast system updates. Having said that, once you make the migration, there are lots of benefits. That is why financial institutions are transitioning customer data to cloud-native. Cloud-based solutions accelerate time-to-value, reduce the cost of acquisition, and enable companies to increase revenues quickly, without having to worry about the maintenance of additional infrastructure. I would need scientists to work my alerts as regular investigators would struggle. The contrary is true. Results

3 ways AI can help prevent AML compliance fines in 2022

January 13, 2022
2021 was another bumper year for fines slapped against financial institutions (FIs) for failures in anti-money laundering (AML) compliance. AML shortcomings in transaction monitoring are a global problem. Countries whose banks were hit with fines include the United States, Germany, the Netherlands, Norway, Latvia, France, the UAE, India, Malaysia, and South Africa.  Fines imposed on FIs by regulators could reach as high as $2 billion for a second year running when the final figures come in, according to estimates. The continuous vigilance of regulators should serve as a wake-up call for financial institutions worldwide to take stock in failures and take action to change the trend in 2022. Some guilty parties lacked an AML compliance culture or even engaged in outright fraud and corruption. Others turned a blind eye. For FIs investing in large and costly compliance teams and tools, it’s surely frustrating to be hit with fines of tens of millions of dollars for non-compliance.  Some banks fined in 2021 were faulted with poor AML programs including implementation or operating outdated AML systems. As we know well, weak AML controls can open the door for financial crimes. With the rising number of players in the financial ecosystem and the growing volume of global cross-border payments, the challenges are not going to diminish. Just the opposite. Financial institutions are surely asking themselves what do to next. Thanks to the advancement of AML technologies based on AI, a direct path has been paved to reverse the trend.  Indeed, 2022 can be the year of change by bringing AI onboard for AML efforts. Here are some ways banks, payment service providers and fintechs can benefit from AML programs powered by AI.    Reduce false alarms Regulators require financial institutions to submit SARs to cover activity suspected as money laundering, terrorist financing or other criminal offenses such as cybercrime and fraud.  They serve as a tool for regulatory and law enforcement bodies.   Failing to submit SARs is a serious violation of AML laws. Many banks took a hit and were fined for this violation in 2021. FIs operating legacy rules-based systems are inundated with a high volume of false positives, which can reach up to 99% in some cases since they monitor for set thresholds. With so many transactions red-flagged, it’s easy to understand how AML compliance teams would have difficulty deciding which cases warrant referral to financial law enforcement. Besides being expensive to operate as they require huge manpower, known rules can be easily accessed and outsmarted by bad actors including in cyber-engineered attacks.  Using unsupervised machine-learning AI for transaction monitoring can pinpoint abnormal activity outside normal patterns of financial transactions. In this way, FIs can focus on the cases that really matter, reduce investigations, and avoid slowing down and blocking transactions. Semi-supervised AI based on known cases of money-laundering can also further train and perfect the detection system. Fill in missing information Poor monitoring of customer identities was another factor that caused FIs to miss cases of suspected money-laundering.   In

Forces driving demand for RegTech

December 3, 2021
RegTech is getting more attention in the fintech ecosystem this year, alongside the surge in cross-border money transfers and expansion of digital transfer platforms. Accelerated by the COVID-19 pandemic, the popularity of electronic payments is spiking both the volume and complexity of financial transaction data. More sophisticated digital financial crimes are another side effect of the uptick in electronic transfers, with fast, ecommerce transactions a vehicle for criminals to exploit and cover money-laundering and other illicit financial activities. These forces are challenging banks and payment service providers to fulfill their compliance obligation to help weed out the financial crime from normal activity. As a result, financial institutions (FIs) are seeking more technologically advanced solutions to automate transaction monitoring. Regulatory technology companies, known collectively as RegTech, are improving tools to help manage increased volumes with optimized performance. Legacy control systems are limited by rules, when the focus should be on highlighting risk factors. RegTech solutions applying AI and machine-learning technology to big data in cloud environments can help financial institutions and payment service providers ensure their growth trajectory, while enabling a risk-based approach. With enterprise momentum toward cloud-based systems, SaaS cloud-hosted RegTech solutions can especially streamline operations. Here’s a deeper look at three major forces driving demand for RegTech solutions:   More action required With the rise in financial crimes, regulators globally are demanding more action to combat money-laundering. Some examples of the many recent directives from regulators around the world:   In the UAE, new guidelines by the central bank from November reflect Financial Action Task Force (FATF) standards. Money transfer firms must maintain effective AML/CFT programs using a risk-based approach to fulfilling obligations, including strong customer due diligence, continuous transaction monitoring, and suspicious transaction reporting. In Asia, the Hong Kong Monetary Authority (HKMA) is specifically encouraging the adoption of cloud-based RegTech solutions as, more advanced technologies can “identify high-risk relationships, suspicious transactions and networks of mule accounts.” The EU is also moving to crack down on money-laundering, kicking off a major campaign headlined: #EUstopsdirtymoney. A new EU AML authority will be set up requiring tighter controls across the EU to ensure the private sector consistently applies AML rules and regulations. Meanwhile, the Bank of Ireland issued new guidelines in June, in line with the European Banking Authority, highlighting the issue of de-risking, noting it is not acceptable for firms to terminate large categories of customers without conducting individual risk assessments to determine whether there are any increased CDD measures which could be applied to allow the customer relationship to be maintained. Rising costs of compliance High costs of compliance are also driving FIs to search for new and more efficient solutions from RegTech providers. The operation of compliance departments and manpower involved in investigations can weigh heavily on the bottom line of FIs. Rules-based systems can be counterproductive and ineffective in dealing with the rising volume of data, as they are known for triggering a high rate of false positives and a large volume of alerts. As a result,

How to ride through the holiday season without too many obstacles

November 22, 2021
Black Friday, Cyber Monday, Cyber Week, Black November, Christmas and New Year’s vacations, aka the commercial part of the “Holiday Season.” For shoppers, retailers, and holidaymakers, it’s one of the most satisfying times of the year. For banks and payment service providers, however, the surge in spending and transactions that comes with this peak season can easily turn into an annual headache. The holiday season is the time of the year when financial criminals can more easily take cover in the surge in global transactions to engage in money-laundering and other illicit activities. Cybercriminals are experts at impersonating people and organizations, which can pose an extra risk when consumers and do-gooders can be off guard as they relish in the frenzy, are feeling generous and more open to special offers. To tackle the uptick in financial crime as transactions surge over the holidays, banks and payment service providers might ramp up compliance operations to support the increased data coming in from higher than usual money-transfer volumes, including both cross-border payments and domestic transfers. But it’s not the only way to head off the compliance challenges posed by these seasonal fluctuations. Here are some ways to better prepare for any eventuality and make it through the holiday season with your resources and avoid trouble from the regulators. Don’t get caught relying on last year’s cases As we have seen since the beginning of the pandemic and surge in digital payments, reality changes very fast. Financial criminals change direction constantly to avoid getting caught. Realistically, it’s therefore not possible to rely only on what you know from the past. Rules-only based transaction monitoring systems are therefore inherently outdated. They can’t be prepared for what new typologies might hit today or tomorrow. When the holiday surge hits and something new surfaces, there won’t be enough time to scramble and write new rules for the system to implement. The effort needed to investigate and analyze during and after a surge period can therefore impede on the capability to file SARs on time, exposing banks and financial institutions to higher costs fines. Detect the anomalies in the crowd Instead of telling the system what to look for like in a rules-only based method, machine-learning- based monitoring using AI can detect previously unknown types of irregularities. An AI-driven approach can manage the unknowns intuitively, calling out something that is out of the norm without knowing its first name. Machine learning lets the data lead the way to the abnormalities and things that are different, saving on the need for analysts to investigate huge volumes of false alerts amid surging transactions. Benefits include: Avoid slowing down and blocking transactions Reduce false positive number and investigation time significantly Identify unknown cases in projects As a result, analysts and compliance officers can benefit from a type of crystal ball to see what is happening in the transaction world irrespective of the number of interactions taking place when a surge takes place. Don’t let your data have limits During transaction surge periods,