Over 800 million people use mobile applications each day, with an increasing amount relying on mobile banking apps. With the rate of mobile banking adoption having grown in recent years, mobile device fraud has inevitably increased as well. Therefore, the need to detect fraudulent transactions has become all the more important.
In a blog post, Microsoft has now explained how to create and implement an AI-based model in Azure which reduces fraud detection times to less than two seconds. This enables customers to take much quicker action against any such activities, improving the security of their belongings.
According to the tech giant, most mobile frauds are initiated through a technique known as a "SIM swap attack". Through these attacks, mobiles numbers are hacked and cloned, with the hacker receiving all SMS notifications and calls destined for the original device. Login credentials can then be obtained through a variety of hacking techniques, eventually leading to the criminal gaining control over a customer's banking details, and hence, funding transfers and withdrawals via mobile.
To mitigate these attacks, Microsoft has created a new fraud detection solution that can flag a fraudulent transaction within two seconds. Many banks are already using similar solutions with a number of other approaches in order to reduce their losses. However, the Redmond giant's new Azure-based model is different in that it critically reduces response times, and offers increased detection rates. Using this solution, incoming mobile activity will be processed, a behavioral profile will be built, transactions will be evaluated for fraud, and relevant action following this process will be determined - all in the space of less than two seconds.
The architecture around which this model is formed involves three major components: Azure Functions, Azure SQL, and Azure Machine Learning. Furthermore, the mechanism detailed in this solution is based on three factors:
- Feature engineering to create customer and account profiles.
- Azure Machine Learning to create a fraud classification model.
- Azure PaaS services for real-time event processing and end-to-end workflow.
Kate Baroni, Principal Software Developer, Azure Engineering, has written the "Mobile bank fraud solution guide" with the express purpose of explaining how the firm's new AI-based model can be implemented. If you're interested in obtaining further information regarding this fraud detection solution, you can download the guide here.