# Korea Insurance Institute Rebuts Claims Over AI Joint Venture > Source: > Published: 2026-08-12 06:07:03+00:00 # Korea Insurance Institute AI Venture Draws Scrutiny On August 11, the Korea Insurance Institute announced plans to use AI for all questions in an upcoming insurance certification exam, while its proposed AI development joint venture faced criticism from industry officials and its labor union. Asiae reported that President Ha Tae-kyung publicly criticized the Financial Services Commission for opposing the venture, which the union said he was pursuing unilaterally. On August 11, the Korea Insurance Institute announced plans to use AI to generate 100% of the questions for an upcoming insurance certification exam, while its proposed AI development joint venture faced criticism from insurance industry officials and the institute's labor union. Asiae reported that the union objected to what it characterized as President Ha Tae-kyung's unilateral effort to establish an AI-related subsidiary. The dispute has also involved the Financial Services Commission. Asiae reported that Ha publicly criticized the FSC for opposing the proposed AI development joint venture. ### Exam-generation system The controversy comes as the Korea Insurance Training Institute, identified in the available reporting as the organization led by Ha, prepares to use an AI-based test-item generation system for insurance certification exams. Asiae reported on August 11 that the institute intended to use AI to generate 100% of the questions for the August Certified Full Sales Competency Exam. According to Asiae, the system was completed in April and its patent was registered in July. The Seoul Economic Daily likewise reported that the institute developed the insurance-sector question-writing system in April and had completed patent registration. An institute representative told Asiae that AI-generated questions reduce the time and cost of test development compared with expert-authored questions and can increase question diversity. Ha said the patent registration was an achievement in the institute's AI-driven education and assessment work, adding that the organization would pursue external commercialization of AI-based test-generation technology. ### Governance questions around institutional AI ventures The reporting does not establish the venture's ownership structure, financing, or technical scope. It does, however, place an assessment-automation effort alongside a dispute over whether a nonprofit training body should create or invest in an AI venture. Across regulated industries, comparable efforts often raise governance questions distinct from model quality: who owns resulting intellectual property, how commercial activity is separated from nonprofit functions, and what oversight applies when institutional data or domain expertise supports a new entity. For ML practitioners, exam-item generation also requires validation processes for factual correctness, curriculum coverage, duplicate or near-duplicate questions, and potentially biased outputs before high-stakes assessment use. ## Key Points - 1The proposed AI joint venture faces regulatory and labor scrutiny, adding governance risk beyond the underlying technology deployment. - 2The institute announced plans for full AI generation of questions for an insurance certification exam, making quality assurance central to assessment reliability. - 3Comparable regulated-sector AI ventures commonly require clear oversight for intellectual property, commercialization, data use, and validation. ## Scoring Rationale The story combines a concrete AI assessment initiative with a governance dispute over an AI-related venture at an insurance-sector institution. It is relevant to practitioners building high-stakes evaluation systems, but the available reporting provides limited technical detail and the event is primarily regional. ## Sources Public references used for this report. Practice with real Health & Insurance data 90 SQL & Python problems · 15 industry datasets 250 free problems · No credit card [See all Health & Insurance problems](/problems/datasets/health)