LG EXAONE to Handle Over 10 Million Safety Reports Annually

Selected as the “Brain” of the Ministry of the Interior and Safety’s AI Safety Sinmungo
EXAONE 4.5 to Launch Pilot Service Within the Year

On April 30, LG AI Research announced that it has completed the first phase of research and development for the Ministry of the Interior and Safety’s ‘AI Safety Sinmungo’ project, in collaboration with the Korea Electronics Technology Institute (KETI), based on LG’s artificial intelligence (AI) model ‘EXAONE.’ The organizations have begun preparing for a pilot service launch within this year.

EXAONE has been designated as the “brain” of the ‘AI Safety Sinmungo,’ a core initiative under the government’s policy agenda to “establish a disaster safety management system ensuring public safety.” The ‘AI Safety Sinmungo’ is a next-generation safety report processing system in which AI analyzes more than 39,000 safety reports submitted daily, and automates the entire process—from report intake and screening to categorization, transfer to responsible departments, and response delivery.

Currently, the Safety Sinmungo employs a partially keyword-based automatic classification system. However, when reports contain typos or ambiguous wording, accuracy declines, requiring staff to manually review attached photos or videos and then classify and forward the reports to the appropriate agencies. The vision-language model ‘EXAONE 4.5,’ which is capable of understanding and reasoning with visual information, serves as the brain to accurately screen and categorize safety reports that include photos and videos.

KETI plans to use the more precise reporting language generated by ‘EXAONE 4.5’ to identify and prioritize critical types of safety reports, thereby shortening the handling period of important cases and reducing administrative workload. For example, if a citizen uploads a photo of a blocked storm drain to the Safety Sinmungo, EXAONE analyzes the image and automatically generates the report details. Reports are then categorized by type, and high-priority cases, such as blocked storm drains during the rainy season, can be transferred directly to the response department, bypassing the classification department, to enable swift resolution.

The two institutions expect that as more safety report data accumulates, it will become possible to analyze patterns based on time, region, type, and frequency. This data-driven analysis can be used to inform policy-making processes that proactively address new types of safety risks.

Shin Heedong, President of KETI, stated, “Through our collaboration with LG, we expect to reduce the administrative workload and operating costs of public services, as well as contribute to enhanced citizen safety,” adding, “We aim to create a model case for AI transformation that the public can tangibly experience by processing safety inquiries quickly and accurately.”

Im Woohyung, Co-Director of LG AI Research, emphasized, “LG’s goal is to generate real value through expert AI,” and added, “By utilizing EXAONE to improve both the speed and quality of safety administration directly related to public safety, we hope to contribute to a better quality of life.”

LG AI Research is leading the AX (AI transformation) business based on EXAONE, in partnership with LG CNS, which handles LG’s public sector projects. LG CNS is leveraging EXAONE in large-scale public AX projects, collaborating with the Ministry of Foreign Affairs, the National Police Agency, and Gyeonggi Provincial Office of Education to apply AI in on-site work.

Given the sensitive nature of information handled in the public sector, LG AI Research has established a dedicated AI ethics organization. By building a proactive management system centered around this organization, the company is working to prevent risks and is developing a ‘Responsible AI’ standard across all stages—from planning and data collection to training, AI model deployment, service operation, and real-world application.

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