As new forms of insurance fraud utilizing generative artificial intelligence (AI) for forgery of medical certificates and receipts proliferate, financial authorities are moving to establish a unified detection system that links data from the insurance sector and public institutions. Unlike traditional forgery methods using Photoshop and manual alterations, AI-generated forgeries leave fewer traces, making them harder to detect. This initiative marks the beginning of a comprehensive government response aimed at preventing insurance losses and the deterioration of health insurance finances.
On June 4, the Financial Services Commission held a kickoff meeting for the "AI-Based Insurance Fraud Prevention Task Force (TF)" alongside an insurance investigation council. The financial authorities plan to operate the TF for three months, with the goal of preparing a plan for an "AI-Based Insurance Fraud Prevention System" by September. Following this, they intend to implement legislative amendments and enhance platforms starting in October.
Last year, the amount of detected insurance fraud was reported at 1.1571 trillion won. Considering undetected fraud, the total is estimated to reach approximately 9 trillion won. By category, long-term damage insurance, including health insurance, accounted for 44.7%, followed by auto insurance at 22.4% and life insurance at 21.8%.
Recently, there have been cases where generative AI and deepfake technology have been misused to forge identification documents, medical certificates, and photos of vehicle damage throughout the entire insurance process, from enrollment to claims. A 20-year-old individual in Busan was sentenced to two years in prison for forging a hospitalization confirmation letter using generative AI, defrauding a total of 150 million won from 11 insurance companies.
Concerns are high regarding AI forgery, as it is difficult to detect with existing methods. In the past, when receipts or medical records were cut and pasted or manipulated using Photoshop, there were often clues left, such as changes in font or spacing. In contrast, generative AI can create new image pixels, eliminating traces of manipulation.
The TF will discuss establishing legal grounds for the centralized sharing of information related to insurance fraud, real-time information sharing among relevant agencies, and the analysis of AI-based fraud patterns and risk indices. Plans are also in place to enhance the "AI-Based InsurTech Platform" of the Korea Credit Information Corporation as a unified infrastructure for preventing insurance fraud across all insurance sectors and to activate a system for original data verification with public institutions.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.