Danske Bank Utilises AI to Enhance Fraud Detection

Summary:

Facing a low 40% fraud detection rate, Danske Bank sought to revamp its fraud detection mechanisms.

Through the implementation of a modern enterprise analytic solution utilizing AI, Danske Bank achieved success.

Each AI model processes real-time data, continually learning to identify traits indicative of fraud. As a result, the bank experienced a notable 60% decrease in false positives.

General:

Danske Bank, a leading European financial institution, implemented AI technology to enhance their fraud efforts. The system driven by AI helped in identifying suspicious transactions with greater precision and speed, consequently diminishing the total duration required for review.

Problem Statement:

Danske Bank, a Nordic universal bank with 145 years of experience in the market, needed to modernise its fraud detection defences as they were struggling with a low 40% fraud detection rate. Danske Bank faced a daily influx of 1,200 false positives, with 99.5% of these cases unrelated to fraud. In a bid to enhance compliance and risk management procedures, Danske Bank opted to strategically implement innovative analytic techniques, including AI.

Results:

1. A significant improvement in compliance and risk management processes.
2. A 60 % reduction in false positives, with an expectation to reach as high as 80 percent.
3. A 50 % increase in true positives.
4. Focusing resources on actual cases of fraud.
5. A notable decrease in fraudulent activities.
6. Enhanced job satisfaction and productivity of the bank’s compliance officers.

AI Solution Overview:

Collaborating with Teradata Consulting, Danske Bank strategically chose to utilise AI to enhance fraud detection while minimising false positives. A Teradata company has developed an AI-powered fraud detection platform. This engine utilises machine learning to examine tens of thousands of hidden characteristics, evaluating millions of online banking transactions in real-time to deliver actionable insights on both genuine and fraudulent activities.

The fraud detection system employs a ‘champion/challenger’ approach for anomaly detection. In this methodology, each model (referred to as a challenger) learns specific transaction traits associated with fraud and receives additional data, such as customer location, to enhance its accuracy. As a challenger outperforms other models, it earns the title of ‘champion’ and contributes to the training of other models. Implementing a contemporary enterprise analytic solution empowered by AI proved highly beneficial for Danske Bank. The integration of ML algorithms enabled Danske Bank to significantly enhance fraud detection accuracy while decreasing false positive occurrences.

References:

1. Empowering Compliance: AI Solutions Redefine AML Investigations (2024) https://financialcrimeacademy.org/ai-solutions-for-aml-investigations/#:~:text=Danske%20Bank%3A%20Danske%20Bank%2C%20a,reducing%20the%20overall%20review%20time.
2. Danske Bank Fights Fraud with Deep Learning and AI (2019) https://assets.teradata.com/resourceCenter/downloads/CaseStudies/CaseStudy_EB9821_Danske_Bank_Fights_Fraud.pdf
3. How Can Artificial Intelligence Assist In Detecting Money Laundering And Terrorist Financing In Financial Institutions? (2023) https://www.marketscreener.com/quote/stock/DANSKE-BANK-A-S-1412871/news/How-Can-Artificial-Intelligence-Assist-In-Detecting-Money-Laundering-And-Terrorist-Financing-In-Fina-44576196/
4. How Does AI Help in Financial Fraud Detection and Protection. https://eurisko.net/how-does-ai-help-in-financial-fraud-detection-and-protection/

Industry: Financial Institutions