**KT’s AI-Based Voice Phishing Detection Service Prevents Approximately 16 Billion KRW in Potential Losses, Detecting 392 High-Risk Numbers**
Since its launch in January, KT’s AI-based voice phishing detection service has proven its high accuracy, garnering attention. This technology, in collaboration with K-Bank, is set to become the first real-time voice phishing response used in the financial sector.
On the 7th, KT announced the results of analyzing operational data over two months since the commercial launch of the AI-based voice phishing detection service.
According to the analysis, out of 1,528 detectable voice phishing calls identified as “caution” or “danger” grades, the detection accuracy was 90.3%. Among these, 392 calls (25%) were confirmed to be cases on the National Police Agency’s voice phishing blacklist or impersonations of the prosecution or police, proving the effectiveness of the alert function in preventing actual damage.
Based on the government-announced average damage per voice phishing case in 2024 (about 41 million KRW), KT estimated that approximately 16 billion KRW worth of losses were prevented.
However, some false positive cases occurred in normal calls related to billing or payment. KT plans to enhance the AI engine and update the whitelist to improve detection accuracy within the first half of the year.
Furthermore, KT worked with the Seoul Metropolitan Police Agency to link its self-developed voice phishing detection technology with malicious app installation URL information secured by the police, contributing to the prevention of actual phishing damages. They provided the police with 3,667 cases of suspicious numbers in contact with these URLs, allowing the police to prevent damage through direct contact with potential victims.
In this process, measures were taken to freeze accounts for victims of phishing scams that impersonated card deliveries, and a potential victim who was about to hand over 100 million KRW in checks to a scammer impersonating the prosecution was also stopped on-site, preventing a total of 2.01 billion KRW in damage over two months (February to March).
Leveraging these achievements, from April, KT will collaborate with K-Bank to introduce AI-based real-time voice phishing detection technology in the financial sector for the first time. This will enable direct financial fraud responses such as fund withdrawal suspensions.
This service, integrated with K-Bank’s fraud prevention system, will provide real-time detection information to K-Bank when a customer receives a suspicious call, thereby enhancing the effectiveness of fraud prevention and potentially expanding to other financial institutions in the future.
Lee Byung-moo, Executive of KT’s AX Innovation Support Division, stated, “It is encouraging that KT’s AI technology is contributing to actual voice phishing prevention. We hope that more citizens will preemptively block phishing damages through the free ‘Whoscall’ app, available regardless of the telecom carrier.”