Fraud Detection Gap Between Large and Small Financial Institutions Warned by US Treasury
The United States Department of the Treasury has issued a warning about a growing disparity in fraudulent transactions detection capabilities between large and small financial institutions in the Northern Mariana Islands.
The Disparity
According to a recent report, smaller financial institutions lack the necessary data sets to effectively utilize AI-powered detection tools. This leaves them vulnerable to fraudulent activities. On the other hand, larger institutions have access to more resources and data to train their AI models, making it easier for them to detect fraud.
The Solution
The report suggests that the US Department of Finance may contribute its own historical data to help bridge this gap and provide small institutions with access to more robust fraud detection capabilities. Additionally, a centralized clearinghouse for fraud data would allow financial institutions of all sizes to share information and support each other in detecting fraudulent transactions.
Promising Results
Despite the challenges, AI-powered fraud detection has shown promising results. One large institution reported a 50% reduction in fraud after training its AI model on internal historical data. The US Department of Finance itself has recovered $375 million through the use of AI-powered check fraud mitigation methods.
New Risks Introduced by AI
The report also warns financial institutions to be aware of new third-party risks introduced by AI, such as generative AI and deepfakes, which can be used to create sophisticated identity impersonation techniques. To mitigate these risks, the report calls on financial institutions to:
- Affix digital labels to vendor AI systems and third-party data
- Provide transparency over how the models were trained and what data was used
Explainability and Fairness
The use of AI in cases that raise concerns over privacy and consumer protection will also require a higher level of explainability. Additional research and development are needed to ensure fairness and bias are not introduced into AI decision-making processes.
By addressing these challenges, financial institutions can improve their fraud detection capabilities and better protect their customers from fraudulent activities.