LG CNS将为韩国金融监督院建设AI不公平交易调查系统，但合同金额较小

South Korea's financial regulator will begin full operation in 2028 of an artificial intelligence model that detects and investigates unfair trading in capital markets, including stock price manipulation. The AI will identify abnormal trades and linked accounts and propose the additional evidence needed for further investigation. A separate AI agent will also be built to support administrative tasks such as drafting investigation reports. LG CNS (064400), the IT services affiliate of LG Group, will develop the system.
LG CNS was recently selected as the preferred bidder for the Financial Supervisory Service's project to build an AI-based system for responding to unfair trading, according to IT industry sources on the 30th. The project, worth about 4.5 billion won, covers development of an AI model specialized for capital market investigations along with an investigation system equipped with various AI functions. LG CNS won the preferred bidder position after competing with Bespin Global and Konan Technology (402030).
At the core of the project is using AI to detect suspected unfair trading early — including leveraged buyouts financed without capital and stock price manipulation — and to raise investigative efficiency. To that end, an AI model specialized for capital market investigation work will be developed. LG CNS plans to build the dedicated model by fine-tuning an existing publicly available general-purpose AI model for the FSS's work.
The model will learn not only from public information such as share prices, trading volumes and regulatory filings, but also from non-public material the FSS has obtained during investigations. Records of past cases accumulated at the FSS, including investigation plans, reports and disposition opinions, will also serve as key training data.
On that basis, the AI model will detect trades suspected of unfair trading and automatically map the relationships among multiple accounts. It will group suspicious accounts to analyze trading circumstances and suggest to investigators the additional material needed to establish charges. When FSS staff obtain the material the AI has proposed and feed it back into the system, the AI reanalyzes the likelihood of violations based on that input. The result is a "circular investigation system" in which humans and AI repeat data collection and analysis to sharpen investigative precision.
Because the system handles sensitive investigative information, the AI model will be installed on a closed network separated from the external internet. The FSS barred the use of services that call models through external clouds, such as OpenAI's ChatGPT or Google's Gemini, for this project. As a result, LG CNS's proposal is understood to include two domestic large language models as candidates, among them the LG AI Research's EXAONE, whose model weights are publicly available and which can therefore be installed on in-house servers.
Another pillar of the project is the development of AI agents to assist FSS staff. The AI agent for unfair trading investigations will be provided primarily in chatbot form. When a user asks in natural language about suspected unfair trading as of the present moment, the agent will respond based on real-time information. It will also draft reports such as question-and-answer records and disposition opinions in FSS formats, using material gathered by investigators. All AI agents will be embedded in the user interface of the FSS's existing work systems. A system to assess the priority of tips and complaints, along with a system for automatically transcribing and summarizing recordings, will also be installed.
The FSS and LG CNS plan to complete the unfair trading investigation AI system by December next year and put it into full practical use from 2028. "We will mobilize all of our expertise in financial services along with our AI model, platform and IT service capabilities to build an AI that contributes to innovating the FSS's response to unfair trading," an LG CNS official said.