Continued Proactive Efforts to Establish AI Ethics
Refining "AI Governance" Through a Systematic Institutional Foundation
KakaoBank is strengthening its AI ethics initiatives across the board to ensure the responsible use and ethical operation of AI technology.
In 2023, KakaoBank became the first financial institution in Korea to obtain the "Artificial Intelligence Management System Certification (ISO/IEC 42001)," which recognized the stability, ethical standards, and efficiency of its AI management system. In 2024, the bank began full-scale operation of its AI governance framework, establishing a trust-based foundation for all AI-related decision-making, including the roles and responsibilities of the organization, ethical principles, leadership, risk level assessment, and compliance evaluations at each stage of the AI lifecycle.
In addition, KakaoBank is accelerating the implementation of more concrete AI ethics practices, encompassing both institutional and technical aspects, such as strengthening internal governance and ensuring technological safety and transparency.
To further advance the "AI Governance" framework, which was successfully established last year, KakaoBank launched the "AI Governance 2.0" project this year.
"AI Governance 2.0" is a strategic initiative to further refine and expand the bank’s governance system, building on KakaoBank’s experience as a leader in AI governance within the financial sector.
As part of this effort, KakaoBank has identified and is proactively implementing key action items in core areas, including: compliance (responding to the revision of the "AI Guidelines for the Financial Sector" and the enforcement of the "Basic AI Act"); policy (expanding the scope of AI governance); organization (considering the establishment of additional AI ethics-related teams); and process (improving compliance procedures throughout the AI lifecycle).
From a technological perspective, KakaoBank is also actively engaged in research and practical application to enhance the transparency and reliability of AI.
KakaoBank is applying "Explainable AI (XAI)" technology to its internal analytics systems, such as its Fraud Detection System (FDS). XAI is a technology that clearly presents users with the operational principles and rationale behind specific AI-generated results, thereby increasing accountability and interpretability. Through this, KakaoBank is improving the accuracy of AI-based operations and strengthening trust across its services.
In addition, the bank has entered into an industry-academia partnership with the KAIST Center for Explainable Artificial Intelligence (XAIC) to jointly develop XAI algorithms and expand the potential for applying these technologies in various financial fields.
Furthermore, KakaoBank has independently developed and operates its own large language model (LLM) evaluation system called "DUO." By utilizing DUO, the bank can quickly and accurately assess models that meet its needs in areas such as financial dialogue capabilities and computational skills among various LLMs. The system also allows for the evaluation of bias and hallucination potential based on ethical, safety, and truthfulness criteria. In this way, KakaoBank uses DUO to select highly reliable AI models suitable for the characteristics of the financial industry and thoroughly reviews their applicability to actual financial services.
KakaoBank plans to continue strengthening both its institutional foundation and technological execution to create an environment in which financial consumers can safely use AI-based services.
KakaoBank stated, "AI ethics is an essential element that must accompany technological advancement, and it is important to put this into practice through both institutional and technological means. KakaoBank will continue to earn customer trust through comprehensive lifecycle management and responsible operation of AI specialized for the financial sector."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
