The FSC introduced a verification system for artificial intelligence (AI)-driven credit scoring model and a security guideline for the use of AI in the financial sector on April 17 with an aim to create a more credible environment for AI usage. These are follow-up measures for the plans to promote AI usage and ensure credibility in the financial sector announced in August 2022.
Key Details
I. Verification System for AI-driven Credit Scoring Model
The verification system will examine whether credit bureaus (CBs) are properly managing credit data in line with peculiar characteristics of AI, whether these companies have made a reasonable selection for algorithms and variables used in their credit scoring models and whether the credit scoring models developed by these companies are statistically significant. The verification system will also look into whether CBs can adequately explain to financial consumers about their credit scoring models and credit assessment results.
The verification committee on personal credit scoring system will make use of this verification system to examine CBs (offering credit scoring service to sole proprietors) that are using an AI-based credit scoring model within this year. Examination of the AI-based credit scoring models used by CBs offering personal credit scoring service to individuals will also take place in the future.
II. Security Guideline for Use of AI in Financial Sector
The security guideline for AI usage in the financial sector offers stage-by-stage security issues that need to be considered when developing an AI-based model while providing a security checklist for AI chatbots. First, the state-by-stage security issues that need to be considered are organized in the order of (a) collecting data for machine learning, (b) preprocessing of data, (c) designing and learning AI models and (d) verifying and assessing AI models. The security guideline provides suggestions on the data management and processing method, model designing technique and security verification method for countering specific security threats such as a contamination of data, personal information leak or attack on AI models. Second, the security guideline provides a separate security checklist for AI chatbots where AI is most widely adopted in financial services. The security checklist is expected to help field practitioners to more easily check security-related matters in a simple checklist format.
The security guideline for AI usage in financial sectors is currently posted on the financial security regtech portal (regtech.fsec.or.kr), and it will be improved and upgraded on a continuing basis taking into account newly emerging types of security threats and counterstrategies.
The FSC stated that AI is an essential technology for developing innovative services in an era of hyper-intelligence, hyper-connectivity and hyper-convergence, and that the financial sector has significant potential for AI usage due to abundance of high-quality data. In this regard, the FSC said that authorities will work to effectively implement the measures included in the plans to promote AI usage and ensure credibility in the financial sector announced in August 2022.
* Please refer to the attached PDF for details.