[Edaily Reporter Kim Se-yeon ] As the use of generative artificial intelligence (AI) rapidly expands across various industries, the insurance sector has reached a point where AI is now involved in the claims review process. Moving beyond the role of an “assistant” that answers customer inquiries or organizes documents, AI is now being actively utilized to determine whether to accept insurance applications or pay out claims. Experts note, “As AI becomes deeply embedded in insurers’ decision-making processes, concerns about judgment errors have also grown,” adding, “Clarifying liability and establishing consumer protection measures are emerging as new challenges.”

◇AI Automatically Reviews Over 50% of Insurance Contracts
An analysis of the Environmental, Social, and Governance (ESG) reports from major property and casualty insurers on the 11th revealed that SamsungFire&MarineInsurance (#SamsungFire&MarineInsurance), Hyundai MARINE & FIRE INSURANCE CO.,LTD(001450), and DB INSURANCE(005830) have expanded the scope of AI application from customer service to policy underwriting, claims review, and accident compensation.
Hyundai MARINE & FIRE INSURANCE CO.,LTD announced that, as of the end of last year, it had automatically underwritten 52.3% of all insurance policies using its AI-based automated underwriting system, “2Q-PASS.” This means the company is utilizing AI not only for simple document review but also in the process of determining whether to accept a policy. Its AI voice bot for auto claims has also expanded its functions from simply guiding customers through accident reporting to providing information on repair costs and estimated insurance payouts.
DB INSURANCE has introduced AI-based pre-underwriting to significantly reduce underwriting time before policy issuance. Tasks that previously took an average of about 40 minutes—such as entering basic information and verifying medical disclosures—have been reduced to approximately 10 minutes since the AI was implemented. The company laid the groundwork for automation by establishing an AI assistant system in February 2024 and obtaining a patent for an AI-powered automated fault determination system in April of last year.
SamsungFire&MarineInsurance is also utilizing medical underwriting AI to analyze medical certificates and surgical records and make initial determinations on whether to pay insurance claims. The company has expanded the scope of AI utilization by introducing AI advisors to its automotive breakdown assistance service.
◇Insurance Companies Boosting Productivity Through Expanded AI Underwriting
The insurance industry is expanding the scope of AI adoption because of its significant impact on improving productivity. Underwriting and claims review are prime examples of labor-intensive tasks that require reviewing large volumes of documents and medical records. Having AI handle repetitive reviews offers the advantage of speeding up the review process, reducing costs, and ensuring consistent standards are applied.
The industry is already achieving tangible financial results through the use of AI in simple, repetitive tasks such as data entry and document verification. DB INSURANCE, for example, automated 216 standardized tasks, saving over 65,000 hours of work annually and achieving direct cost savings of approximately 1.5 billion won.
However, the expansion of AI use presents significant challenges. Since decisions regarding insurance application rejections or claims payments are directly linked to consumer rights, it remains unclear who would be held accountable if AI makes an incorrect judgment, or what procedures consumers can follow to file an appeal and seek redress. Consequently, the pace at which AI is expanding into core decision-making areas is stalling. Ultimately, institutional improvements that clarify explainability and the party bearing ultimate responsibility are necessary to protect consumer rights and enhance operational productivity in the insurance industry.
Song Yoon-ah, a research fellow at the Korea Insurance Research Institute, also noted in a report titled “The AI Implementation Gap in the Insurance Industry” released late last month, “While underwriting decisions (determining whether to approve a policy based on underwriting results) are not tasks that the law mandates must be performed only by individuals with specific qualifications, they carry significant ultimate responsibility and verification burdens.” “Since human managers are required to verify and approve outputs generated by AI, this tends to slow down the pace of technology adoption.” She went on to recommend, “It is necessary to proactively establish governance frameworks to address AI errors, bias, and liability.”