The South Korean government is drawing up a national science-based policing strategy to counter rapidly advancing crimes powered by artificial intelligence. Concerns are mounting that AI is being misused across the entire criminal process — from identifying potential victims to extracting confidential information and money — and not merely to produce deepfake images and voice clones. The government plans to review every research and development project aimed at countering new AI-driven crimes and use the findings to build a medium- to long-term strategy.
According to the Ministry of Science and ICT and the Presidential Advisory Council on Science and Technology on the 7th, the government will soon launch a study to build a science-based policing framework and establish a national strategy. Under Article 33 of the Police Act and regulations on promoting science and technology in policing, the National Police Agency draws up a comprehensive plan for policing technology every five years. Officials concluded that the plan alone has limits in effectively blocking AI-driven crime.
A government official said the police agency's existing plan is a broad document covering both R&D and single-year operational projects, and that the agency's R&D is somewhat weaker than the defense sector in terms of budget and organization. The official added that because new types of crime are becoming more sophisticated and contactless through technologies such as AI and the dark web, there is a need to present a medium- to long-term roadmap focused on R&D. According to the National Police Agency, cases of editing and distributing false video material — deepfakes — under the law on punishing sexual violence surged from 168 in 2023 to 550 in 2024 and 1,300 last year.
The bigger concern is that AI may advance beyond processing specific commands to committing crimes on its own. In a report last year, the United Nations Office on Drugs and Crime said phishing kits packaging malicious websites and email templates that can be deployed with minimal effort are being sold. It said one such kit, "16shop," operated out of Indonesia, was sold on the dark web for about $60, or roughly 80,000 won, leading to more than 70,000 victims in 43 countries and the creation of more than 150,000 phishing domains. WormGPT, FraudGPT and Nytheon AI are also known as software-as-a-service hacking tools optimized for phishing crimes.
Jara Perumal, co-founder and chief technology officer of Overwatch Data, a U.S. firm that uses AI to monitor intelligence and threats, testified before a subcommittee of the U.S. House Judiciary Committee in July last year that agentic AI is becoming a new trend in cybercrime. Unlike conventional bots that followed fixed scripts, the executive warned, an AI agent given only a rough description of a scam can work out the step-by-step process and adjust its methods as needed, carrying out crimes in ways that are harder to detect and longer-lasting than existing systems. The global AI agent market is projected to grow from $7.84 billion last year to $52.62 billion by 2030 — from about 11 trillion won to roughly 71 trillion won — even as the technology is turned to criminal use.
The government will review all R&D projects on new types of crime pursued in South Korea since the emergence of generative AI, including dark web drug investigations and platforms for predicting crimes between people who know each other. It plans to identify the technologies front-line officers need, assess how far those technologies have advanced, and diagnose domestic R&D blind spots based on an analysis of supply and demand. The advisory council aims to use the results to build a full-cycle response and integrated support system spanning prevention, investigation and victim protection, and to establish the national science-based policing strategy in the first half of next year.
The government also plans to consolidate the organizations tasked with responding to AI crime. After launching a cross-ministry consultative body on AI crime in June this year, it is now pushing to create an integrated AI crime response unit to oversee the effort. An official at the Korea Communications Commission said the unit will serve as a control tower for the response to AI crime, drawing up and managing a comprehensive cross-ministry plan, coordinating cooperation and roles among ministries, working with the private sector and activating rapid-response systems when social issues arise.
