{
  "id": 20253646,
  "title": "Shinhan Bank Presents Two Research Papers on Trustworthy Financial AI at Global Conference in Milan",
  "url": "https://urgent.news/2026/11/16/shinhan-bank-presents-two-research-papers-on-trustworthy-financial-ai",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-16T03:52:51.000Z",
  "source": {
    "name": "Korea IT Times",
    "slug": "korea-it-times",
    "url": "https://www.koreaittimes.com/news/articleView.html?idxno=159179"
  },
  "original_language": "en",
  "account": "Shinhan Bank recently shared two research papers on trustworthy AI for financial services at an international conference in Milan. The studies were unveiled on November 15 at the 2026 International Conference on AI in Finance (ICAIF), a gathering of researchers and industry leaders focused on the intersection of finance and artificial intelligence. The papers introduce new frameworks to evaluate AI models and agents in real-world banking scenarios. One framework, called K-BankEval, assesses how well large language models (LLMs) perform specific banking tasks like lending, deposits, and foreign exchange. It consists of 5,000 questions across six categories, including expertise, document retrieval, tool usage, calculations, safety, and task planning. Tests against six LLMs showed that general-purpose AI models do not always excel in banking-specific tasks, underscoring the need for industry-specific evaluation standards. The second paper presents an evaluation framework built on Shinhan Bank's internal Agent Foundry platform. This system not only examines the final responses generated by AI agents but also the steps they take to complete tasks. It converts operational requirements into specific tests and criteria, then analyzes both the responses and the tools used. The framework helps identify where AI meets banking standards and where improvements are needed before deployment. Shinhan Bank aims to enhance the reliability of AI in banking by using these findings to refine their AI-assisted operations. Officials emphasized that while high-performing AI is essential, verifying its suitability for business operations and ensuring compliance with standards is equally important. The bank plans to leverage these insights to build more dependable AI systems for the financial industry.",
  "summary": "Shinhan Bank is advancing efforts to make artificial intelligence more reliable and secure in financial services, presenting two research papers at an international academic conference in Italy.The bank announced that it presented the research on November 15 at a workshop of the 2026 International Conference on AI in Finance (ICAIF), held in Milan. The studies introduce evaluation frameworks…",
  "key_points": [
    "Shinhan Bank presented two research papers on trustworthy AI for finance at Milan conference.",
    "K-BankEval framework assesses LLMs in banking tasks like lending and deposits.",
    "Agent Foundry platform evaluates AI responses and steps for banking compliance."
  ],
  "editors_take": "Developing specialized evaluation frameworks signals a shift toward prioritizing industry-specific compliance and operational reliability over the general-purpose capabilities of standard large language models in banking.",
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}