Anti Money Laundering Algorithms in 2024: Tackling AML with AI
According to the United Nations Office on Drugs and Crime (UNODC), the estimated amount of total money laundered annually around the world is 2-5% of the global GDP ($800 billion – 2 trillion). Legal and financial institutions are adopting software to comply to anti-money-laundering (AML) regulations issued by the financial action task force (FATF) and the international monetary fund (IMF) after 9/11. The market of AML software is projected to reach $1.7 billion in 2023.
What is money laundering?
The United Nations describes money laundering as “the conversion or transfer of property, knowing that such property is derived from any offense(s), for the purpose of concealing or disguising the illicit origin of the property or of assisting any person who is involved in such offense(s) to evade the legal consequences of his actions”
Money laundering typically follows three stages:
- Placement: Placing the funds with legitimately made money via fake invoices, or by physically moving small amounts to off-shore accounts
- Layering: Disguising the trail through a series of transactions and bookkeeping tricks
- Integration: Making the money seem to come from a legitimate source, such as making it seem like a loan from a friend or a salary of an employee
Laundered money could be used to finance terrorism, developing illegal weapons, evade taxes, corruption etc.
What is anti-money-laundering software?
Anti money laundering (AML) refers to the laws, regulations, and procedures which guide financial institutions to prevent, detect and report money laundering activities. AML software rely on analysis of customer data for:
- Customer onboarding: Users can check customers on sanction lists, wanted lists, and politically exposed people (PEP) lists.
- Customer monitoring: AML software can regularly monitor customers’ behavior to detect suspicious transactions.
- Real time transaction monitoring: Every transaction going through financial institutions will be passed through the software for screening. If a party in a financial transaction is determined to belong to a PEP or sanction list, the software generates an alarm and the transaction will be terminated.
- Reporting: AML software automates the creation of Suspicious Activity Reports (SAR), and filing of these reports in order to achieve compliance with AML legislations.
Types of AML software
There are 4 main types of AML software depending on the detection approach:
The goal of the software is to improve the organization’s ability to detect suspicious activities associated with money laundering. It enables:
Currency transaction reporting (CTR)
This software detects currency transactions which typically require reporting (e.g. transactions >$10k in the US). The software scans all bank transactions, then creates and files the reports.
Customer identity management
This software detects suspicious individuals by running customers’ names through various lists including PEP lists and sanction lists. The softwares can check against other databases to provide positive confirmation of ID, such as phone books or post office delivery addresses.
A transaction monitoring software monitors transactions on a daily basis, and tracks customer behavior in order to generate alerts when transactions are outside the normal behavior.
AML software challenges
Some of the challenges that face traditional AML software, especially compliance based and transaction monitoring software, include:
- Data aggregation: Relevant data can exist in various systems. Aggregating and analyzing data in multiple forms (structured and unstructured) from multiple back-end source systems at financial organization can be challenging
- False positives: False positives make up a significant proportion of the alerts that AML monitoring and screening measures generate. For example, banks in China and UK were able to identify that their existing AML transaction monitoring software was producing alerts at a 90-95% false positive rate.
- Risk scoring: Risk scoring protocols are currently based on generic variables about customers such as source of funds, employment, and PEP status. However, if historical data is not used to create the Risk-based Scorecard then a customer may provide false information and obtain a low risk score.
However, many of these challenges can be tackled by AI/ML enabled software.
What are the benefits of applying AI/ML in AML software?
AI and machine learning (ML) anti money laundering algorithms play a vital role in all data management and analytics activities, such that they enable:
- automating investigation processes
- automate document screening using natural language understanding and processing
- data analysis of big databases and resources
- detecting money laundering transactions with machine learning
- detecting potential risks which transaction monitoring systems (TMS) may be missing
- detecting suspicious patterns
- learning from labeled data (e.g. financial crimes)
- regulating suspicious activity reports (SARs)
- reduce false positive reports of money laundering and fraud
What are the challenges of applying AI/ML in AML?
Though AI is a necessity for the growth and improvement in any product today, it is not necessarily easy to integrate into your AML process. Here’s a list of challenges which face AI implementation in AML:
- Data preparation: such as exploring, structuring, cleaning, and shaping can be necessary but these can also be handled by the AI vendor
- Data quality: Machine learning models are as good as their input. If previous money laundering transactions were not identified by the financial institution, then future cases may also be hard to identify. However, a single user’s (e.g. a bank’s) data is not all that power ML models and models trained on diverse, historical data can catch cases that a user was previously not able to catch.
- Continuing role of humans:
- In order to make informed decisions about ambiguous cases a human is required to analyze the data collected and processed by the AI algorithm
- Sometimes the algorithm will face a case on which the algorithm has not been trained before, thus requiring human judgement to avoid error and bias
How can AI/ML tackle false positives in AML?
When an alert of a suspicious activity is generated, it is used as an input to a clustering algorithm which relies on semantic and statistical analyses. The algorithm can:
- identify duplicate or redundant data
- categorize the activity into low, moderate, high, or critical
- detect name patterns which trigger false positive hits on sanctions lists
For example, the HSBC bank utilized Ayasdi AI-enabled AML platform and were able to reduce false positives by 20%. This is a significant improvement but even after this, the bank is still likely to be facing false positives most of the time.
What is the future of AML software?
The need for AML software is growing exponentially due to:
- Exponential growth in data volumes: The total data volume (i.e. not just AML related) was estimated at 64 zettabytes (ZB) in 2020 growing at a CAGR of 23% until 2025 which will require sophisticated AI and big data analytics to handle. This is already evidenced by bank’s increased share of big data and business analytics market. In 2019, banks produced ±14% of big data and business analytics revenues.
- Leveraging machine learning in fraud detection: According to the Association of Certified Fraud Examiners (ACFE), organizations are expected to triple their spending spend on AI and machine learning to prevent online fraud.
- Cryptocurrency money laundering cases: In 2020, global crypto thefts, hacks, and frauds totaled $1.9 billion, and $3.5 billion worth of transactions were associated with criminal bitcoin addresses.
Current AI-powered AML software are able to recognize spending habits, analyze customer behavior, learn criminal patters, as well as create understandable profiles about people and entities. Combined with big data analytics and RPA, we estimate that a great proportion of AML processes can be enhanced by automation.
For more on AI in finance
To learn more about AI/ML and RPA in finance, feel free to read our articles:
- 15+ AI Applications / Use Cases / Examples in Finance
- Top 15 RPA Use Cases in Banking
- KYC Automation: 5 Technologies for Better Transparency
If you are ready to invest in an off-the-shelf AML software, we can help:
- Check our data-driven list of AML solutions
- Reach out to us
Cem has been the principal analyst at AIMultiple since 2017. AIMultiple informs hundreds of thousands of businesses (as per similarWeb) including 60% of Fortune 500 every month.
Cem's work has been cited by leading global publications including Business Insider, Forbes, Washington Post, global firms like Deloitte, HPE, NGOs like World Economic Forum and supranational organizations like European Commission. You can see more reputable companies and media that referenced AIMultiple.
Throughout his career, Cem served as a tech consultant, tech buyer and tech entrepreneur. He advised businesses on their enterprise software, automation, cloud, AI / ML and other technology related decisions at McKinsey & Company and Altman Solon for more than a decade. He also published a McKinsey report on digitalization.
He led technology strategy and procurement of a telco while reporting to the CEO. He has also led commercial growth of deep tech company Hypatos that reached a 7 digit annual recurring revenue and a 9 digit valuation from 0 within 2 years. Cem's work in Hypatos was covered by leading technology publications like TechCrunch and Business Insider.
Cem regularly speaks at international technology conferences. He graduated from Bogazici University as a computer engineer and holds an MBA from Columbia Business School.
To stay up-to-date on B2B tech & accelerate your enterprise:Follow on
Next to Read
Your email address will not be published. All fields are required.