Detecting Altered Voices with AI: KT Launches Real-Time AI Voice Phishing Detection Service 2.0

[Financial News] KT announced on the 29th that it will commercialize the 'AI Voice Phishing Detection Service 2.0', equipped with speaker recognition and deep voice (altered voice) detection functions, starting from the 30th.
This advanced service is characterized by its ability to recognize not only the actual voice of criminals but also identify voices altered by AI, in addition to the existing context-based detection technology.
In the case of speaker recognition, AI precisely analyzes voiceprint information based on actual voice data of voice phishing criminals provided by the National Forensic Service to detect criminal activity. This is a step forward from the existing detection method based on financial keywords or sentence structure. 
Additionally, KT provides a deep voice detection function that distinguishes altered voices generated by AI voice synthesis (TTS) technology. This technology has already been verified through the 'AI Voice Authentication' service, which was first applied to the Health Insurance Corporation's customer center in May last year. KT explained that it is expected to proactively respond to the increasingly intelligent AI-based voice phishing crime methods. 
During the first half of this year, KT's AI Voice Phishing Detection Service analyzed approximately 14.6 million call traffic, recording a detection accuracy of 91.6% and is estimated to have prevented damages of about 71 billion won. Through the release of version 2.0, KT aims to achieve an annual detection accuracy of over 95% and prevent damages exceeding 200 billion won.
Furthermore, KT is working to prevent financial damage by linking voice phishing detection information with the financial sector's Fraud Detection System (FDS). In May, KT signed a Memorandum of Understanding (MOU) with the Korea Federation of Banks to provide detection data to financial institutions, establishing a public-private cooperation system that enables real-time actions such as 'detection → account monitoring → withdrawal blocking'. 
Lee Hyun-seok, head of KT Customer Division, said, "We deeply appreciate the Ministry of Science and ICT, the Personal Information Protection Commission, and the National Forensic Service for their cooperation in commercializing the speaker recognition-based AI detection service to protect citizens more strongly from voice phishing," adding, "With this technology commercialization, we will further strengthen collaboration with the financial sector to drastically reduce customer damage."

psh@fnnews.com Sung-hyun Park, Reporter