본문 바로가기
  • Home

Design of Trustworthiness Evaluation Indicators for Trustworthy Autonomy of AI Agents in the Financial Sector

  • Journal of Internet of Things and Convergence
  • Abbr : JKIOTS
  • 2026, 12(5), 15
  • Publisher : The Korea Internet of Things Society
  • Research Area : Engineering > Computer Science > Internet Information Processing
  • Received : September 23, 2026
  • Accepted : October 15, 2026
  • Published : October 31, 2026

Dongsoo Moon 1

1성균관대학교

Accredited

ABSTRACT

This study aims to design trustworthiness evaluation indicators and a risk-based assessment framework for the trustworthy autonomy of AI agents in the financial sector. We comparatively analyzed major AI governance frameworks, including the EU AI Act, Korea AI Act, Singapore MGF, and NIST AI RMF, to reconstruct six core trustworthiness requirements. By incorporating financial risk characteristics, we derived seven evaluation areas and 14 detailed indicators. We further specified the rationale and measurement methods for each indicator and proposed preliminary assessment criteria for critical safety requirements, including unauthorized execution and approval bypass. A differential assessment matrix was developed based on AI agent autonomy levels (A0~A4) and financial impact levels (F1~F4). The application of the proposed assessment procedure and decision criteria was illustrated through a hypothetical investment order scenario. This study provides a foundational framework for evaluating the safety and controllability of autonomous actions performed by financial AI agents. Further empirical validation is required to establish the validity and effectiveness of the proposed assessment criteria.

Journal Copyright Policy

No CCL information provided

Citation status

* References for papers published after 2025 are currently being built.