A Voice Phishing Detection Solution
While financial crimes through voice phishing are still rampant, Samsung Card has succeeded in developing a detection system for voice phishing using machine learning techniques. Although there has been a fraud detection system (FDS) using artificial intelligence (AI) technology, Samsung Card has become the first company in the domestic card industry to build a system aimed precisely at combating voice phishing using various data and machine learning techniques.
Samsung Card has begun to appy a machine learning-based voice phishing FDS to the credit card loan approval process. The key to this system is to score voice phishing possibilities in real time using various data and machine learning techniques for all transactions. It utilizes not only customer-related information such as income and credit information accumulated at Samsung Card, but real-time transaction information, loans and arrears at other companies and external financial data such as loan demand.
The goal is to detect suspicious transactions and prevent damage to customers before the transactions actually take place. As criminal methods become more sophisticated, the amount of financial damage caused by voice phishing is on the uptick despite crackdowns. According to the National Police Agency, the amount of damage, which averaged 110 billion won per year from 2006 to 2016, surged to 639.8 billion won in 2019. Recently, even voice phishing techniques that take advantage of people through the COVID-19 outbreak have been found.
Samsung Card assembled a project team in November 2019 and finished building the system in February 2020. In a pilot test conducted in March, the system prevented 460 million won in damage. This is the result of applying the system only to the high-risk group which is 0.05 percent of the card loan approval cases. This is equivalent to 40 percent of 1.2 billion won that voice phishing con artists attempted to steal in March. Samsung Card plans to expand the scope of voice phishing detection by applying the system to its cash services.