Pursuing Eight Initiatives, Including a Precision Detection System for AI-Manipulated Content

As crimes involving the misuse of artificial intelligence (AI) technology evolve beyond sex crimes such as deepfakes into various forms, including defamation, investment fraud, and voice phishing, the police are set to upgrade their detection and analysis systems.


Police Begin Upgrading Detection and Analysis Systems to Combat 'AI Misuse' Crimes View original image

The National Office of Investigation of the Korean National Police Agency announced on October 15 that it had established and would implement a master plan to upgrade its system for detecting and analyzing illegally manipulated AI-generated content. The plan aims to proactively respond to crimes involving the misuse of AI and create an AI ecosystem in which the public can feel safe. The police said the plan focuses on enhancing their ability to quickly and accurately detect and determine whether content has been manipulated, and to analyze every stage of crimes involving the use of AI technology to manipulate and distribute content.


Accordingly, the police will pursue eight detailed initiatives across technology, systems, personnel, and organization, including building a precision detection system for AI-manipulated content, developing automated analysis tools and upgrading techniques, and operating AI specialists.


First, the police will build a "next-generation precision detection system" to quickly detect and respond to AI-manipulated content. They have been using a deepfake detection system that is already in use worldwide since introducing it in 2024. They are improving its performance by training it on data tailored to Koreans and adapting the system to conditions in South Korea.


The police are also proactively pursuing a research and development (R&D) project to introduce multimodal deepfake detection technology. Previously, they focused on visual elements such as video and images when examining whether content had been manipulated. Going forward, they plan to improve detection performance by analyzing audio and text together with visual elements. For example, the system will comprehensively assess whether the movements of a person's mouth in a video match the voice, and whether the audio matches the subtitles.


In addition, the police are strengthening analytical techniques for investigating the process of crimes such as creating and distributing manipulated content using generative AI. They are developing methods to analyze the course of a crime by selecting widely used generative AI services and examining usage histories and command logs, identifying specific programs used in manipulation, and analyzing altered data. The police are also distributing relevant guides while conducting research into emerging technologies.


At the same time, the police are developing automated tools to extract evidence by analyzing every stage of crimes involving the creation and distribution of illegal manipulated content using AI technology. To this end, they are also steadily pursuing related research projects, including developing forensic technologies for locally run AI models, with an investment of 10 billion won over four years starting this year.



Acting Korean National Police Agency Commissioner General Kim Jongcheol said, "As crimes involving AI technology are changing rapidly, the police response must change as well. We will equip the police with advanced investigative capabilities to quickly detect AI manipulation and effectively trace not only how it was created but also how it was distributed, so that the public can feel safe."


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing