Enhancing Trust in AI-Based Credit Evaluation... Authorities Establish Verification System

Establishment of an 'AI-Based Credit Evaluation Model Verification System'

On the 9th, officials were busy moving in the corridor of the Financial Services Commission at the Government Seoul Office in Jongno-gu, Seoul, where financial authorities decided to include mortgage loans (Judaemdae) in the 'debt refinancing' infrastructure scheduled to be launched in May by the end of the year. Financial authorities explained that they aim to reduce the interest burden on mortgage loans by building a debt refinancing platform that allows users to compare financial sector loan interest rates at a glance and switch loans easily. Photo by Dongju Yoon doso7@

The financial authorities have established a verification system to enhance the reliability of AI-based credit evaluation models in the financial sector. Additionally, they have developed AI security guidelines to address security threats.

On the 17th, the Financial Services Commission announced the establishment of an ‘AI-based Credit Evaluation Model Verification System’ and ‘AI Security Guidelines for the Financial Sector’ to build a trusted environment for AI utilization. This is a follow-up measure to the ‘Plan to Activate and Secure Trust in AI Utilization in the Financial Sector’ announced by the authorities in August last year.

The AI-based credit evaluation model verification system was established to ensure the reliability of credit evaluation models developed by credit bureaus (CB) using AI methodologies. According to the authorities, AI credit evaluation models can incorporate a wider variety of data compared to traditional models, offering superior predictive power and discrimination ability; however, they have the drawback of using complex algorithms that make intuitive interpretation and explanation of evaluation results somewhat difficult.

Accordingly, through the verification system, the authorities check whether CBs have properly managed credit evaluation models and related data, reasonably selected algorithms and variables, and whether the results are statistically significant. They also verify whether CBs provide sufficient explanations of the credit evaluation models and results to financial consumers.

The Personal Credit Evaluation System Verification Committee of the Korea Credit Information Services will utilize this verification system to conduct verification of CBs using AI credit evaluation models for individual business operators within this year. In the future, verification will also be conducted on AI credit evaluation models of personal credit evaluation companies and others.

Furthermore, the authorities have prepared AI security guidelines for the financial sector. These guidelines present security considerations to be taken into account during the development stages of AI models and provide security checklists for AI chatbot services. The guidelines are posted on the ‘Financial Security RegTech Portal’ and will be continuously improved and supplemented considering emerging security threats.

The financial authorities stated, “AI is a core technology for developing innovative services in the era of super-intelligence, hyper-connectivity, and super-convergence, and the financial sector, rich in high-quality data, has great potential for AI utilization.” They added, “We will continue to steadily implement other detailed tasks of the plan to activate and secure trust in AI utilization in the financial sector.”

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