LG Uplus Launches World’s First On-Device AI to Detect Deepfake Voices (Image supported by ChatGPT)
SEOUL, June 26 (Korea Bizwire) — As voice phishing scams using AI-generated fake voices and images continue to rise in South Korea, LG Uplus is stepping in with cutting-edge countermeasures. The telecom company announced Thursday the commercial rollout of its new ‘Anti-DeepVoice’ and ‘Anti-Deepfake’ technologies aimed at detecting fraudulent audio and video content.
Unveiled earlier this year at MWC 2025, LG Uplus’s Anti-DeepVoice tool has now been embedded into its AI call agent service “ixi-O.” In a world-first, the company has implemented the technology in an “on-device” format, meaning that user data remains securely stored on individual smartphones, enhancing both privacy and speed.
To develop Anti-DeepVoice, LG Uplus trained its AI engine with over 3,000 hours of call data. The system identifies telltale signs of audio manipulation, such as unnatural pronunciations and irregular frequency patterns. It integrates a suite of proprietary tools, including voice activity detection (VAD), speech-to-text conversion, and anti-spoofing analysis.
The tool has been optimized for low-resource environments to ensure it runs seamlessly on mobile devices. Once deployed, ixi-O can instantly detect and warn users of synthetic voices during calls—providing real-time protection against impersonation scams.
In parallel, LG Uplus is developing Anti-Deepfake technology capable of identifying AI-generated facial images and videos. By analyzing pixel textures, frame inconsistencies, and undetectable pattern imbalances, the tool aims to flag forged content even under standard smartphone conditions.
To build a full-spectrum defense against phone-based fraud, LG Uplus is also preparing additional AI features, such as a pre-call phishing detection system that scans high-risk numbers, and a tool that recognizes the voices of known fraudsters. The company is pursuing regulatory sandbox approval from the Ministry of Science and ICT for testing these features.
In collaboration with the financial sector, LG Uplus is also working on real-time AI alerts for suspicious call behaviors, along with tools to counteract digital stalking, verbal abuse, and misinformation during phone conversations.
“We aim to make AI a tool for safety and peace of mind in everyday life,” said Jun Byung-ki, Executive Vice President and Head of the AX Technology Group at LG Uplus. “This is part of our mission to build a brighter, safer world.”
Kevin Lee (kevinlee@koreabizwire.com)
