In 2021 Natwest Group was fined more than £ 264 million after admitting to breaching anti-money laundering regulations. It was the first time a financial institution had faced criminal prosecution by the Financial Conduct Authority (FCA) under anti-money laundering laws in the UK.
Fowler Oldfield, a Bradford jeweler, deposited £365 million with the bank over a five-year-period across multiple branches, including £ 264 million in cash, some of which was brought into a branch in bin bags.
Fowler Oldfield’s predicted annual turnover was £ 15 million when onboarded as a client by NatWest in 2011. But the bank accepted some £ 365 million in deposits over five years from the same client, including £ 264 million in cash. Cash deposits that on occasions came with Scottish notes, which the court heard smelled “musty” as if they had been “stored under the floorboards”.
Rule based transaction monitoring systems, of which FICO was the leader, were largely ineffective with large number of false positives and Compliance organizations only being able to detect 1-2% of the money laundering transactions. Adoption of AI & ML in systems built to detect suspicious transaction activity was very slow. The very next year FICO brought in the AML Threat Score and the AML Soft-Clustering Misalignment Score, one supervised and one unsupervised ML model.
Since then, the use of Machine Learning algorithms to power AML systems that analyze customer data or identify suspicious activities at Banks like HSBC, Standard Chartered and JP Morgan Chase have been on the rise (Artificial Intelligence and Anti-Money Laundering - Sanction Scanner). There is also an increase in the adoption of AI & ML for payments validation and screening to detect frauds across banks, asset managers, insurers and NBFIs (AI Boosting Payments Efficiency & Cutting Fraud | J.P. Morgan).
The likes of NVIDIA have come up with components and capabilities that are optimized to support the retrieval, filtering, scoring and ordering of terabytes of data for banks inclined to building their own monitoring systems. The market has seen newer providers of AI powered niche solutions for customer onboarding with identity confidence and protection against identity fraud.
Going back to the Bradford jeweler, I personally don't think any AI powered transaction monitoring system would have helped Natwest then; not if they were recording direct cash deposits as cheque deposits and 11 internal money laundering reports and 10 automated transaction monitoring alerts didn't result in a single Suspicious Activity Report being filed.
Kommentare