Insurance Research Institute hosts international seminar on emerging risks

Industry shifting from 'AI for insurance' to 'insurance for AI'

Average US data breach cost hits record $10.22 million

Climate risk upends insurance's core principle of risk diversification

Experts call for AI and cybersecurity framework laws, public-private risk sharing

Kim Heon-su, president of the Insurance Research Institute, delivers a welcoming address at the "2026 International Workshop on Risk and Insurance" seminar held Monday morning at FKI Tower in Yeouido, Seoul. The event was co-organized with Pohang University of Science and Technology, the University of Illinois and Simon Fraser University in Canada. [Insurance Research Institute]
Kim Heon-su, president of the Insurance Research Institute, delivers a welcoming address at the "2026 International Workshop on Risk and Insurance" seminar held Monday morning at FKI Tower in Yeouido, Seoul. The event was co-organized with Pohang University of Science and Technology, the University of Illinois and Simon Fraser University in Canada. [Insurance Research Institute]

"It's like handing a child who just learned to read a high school math problem — or giving them a philosophy book and asking for a book report."

That was how Choi Yong-min, a former executive director at Munich Re, described the state of South Korea's cyber insurance market at an international insurance seminar held Monday at FKI Tower in Yeouido, Seoul. Speaking during a panel discussion on AI and cyber risk, he offered a blunt diagnosis: new risks are already at the door, but the industry's readiness to cover them remains at a crawl.

AI-related accidents, increasingly frequent extreme weather events and the compounding shock when such risks collide simultaneously — a new class of hazards is emerging faster than the industry can respond, forcing insurers to rethink what they cover and how far that coverage should extend. The seminar brought together leading international academics to present solutions while also holding up a mirror to where South Korea currently stands.

The seminar, titled "2026 International Workshop on Risk and Insurance," was co-organized by the Insurance Research Institute with Pohang University of Science and Technology, the University of Illinois and Simon Fraser University in Canada. Nine presentations were delivered across three sessions — covering AI innovation and cyber risk, climate change and extreme weather, and insurance data science and market innovation — drawing academics, industry practitioners and policy experts from home and abroad.

From 'AI using insurance' to 'insurance covering AI'

The first session centered on a fundamental shift: from insurers using AI as a tool to AI itself becoming the subject of coverage. Arthur Charpentier, a professor at the Université du Québec, said AI is no longer merely an instrument for insurers but is now embedded in companies' products and decision-making, making it an object of coverage in its own right. The deeper problem, he argued, lies in the nature of the risk: policies that appear diversified across industries and geographies may in fact share dependence on the same AI model or cloud platform, creating the potential for simultaneous, large-scale losses. Riffing on the adage "code is law," he warned that "code written by AI becomes liability."

Alex Jia, a professor at the Geneva Association and Peking University, examined how consumers are changing. A survey he cited found that 68 percent of policyholders had used generative AI tools such as ChatGPT to understand policy terms or compare products. "The era of confusing consumers with complex policy language is over," he said, adding that insurers are evolving from underwriters into service providers. He noted, however, that attitudes toward AI vary by region: Asian consumers tend to be relatively receptive, while European and American consumers more often express discomfort at the disappearing human touch.

Maochao Xu, a professor at Illinois State University, examined why pricing cyber insurance is so difficult. Threats evolve faster than underwriting cycles, loss data is scarce, and rare but catastrophic tail events dominate premium calculations. Last year, the average cost of a single data breach in the United States reached a record $10.22 million (about 15.8 billion won), while the global cyber insurance market has grown to roughly $15 billion (about 23 trillion won).

Domestic experts on the panel offered a grounded assessment. Choi said the industry has moved beyond broad risk categories — such as "a man in his 30s living in Seoul" — to pricing premiums based on individual driving behavior tracked in real time through telematics. He described this "hyper-personalization" as an irreversible trend.

Yet precisely measuring individual risk and covering a new class of catastrophic events are entirely different challenges, he cautioned. Unlike a fire, an AI or cyber catastrophe is nearly impossible to predict in timing or scale. He called for a tiered risk-sharing structure involving private insurers, reinsurers and the government, and urged lawmakers to fast-track a basic AI law and a cybersecurity framework law.

Climate and data redraw the map of risk

The second session focused on the danger of risks arriving in combination. Cihe Tang, a professor at the University of New South Wales, argued that assessing climate shocks and market shocks in isolation leads to an underestimation of potential losses. Only by examining "compound stress" — simultaneous shocks — can the full vulnerability of the financial system be revealed. When new risks interlock and cascade, the damage is amplified.

By the same token, Jose Garrido, a professor at Concordia University, was unequivocal: "Climate risk does not follow the basic principles of insurance." Because it strikes globally and simultaneously, spreading risk across regions becomes nearly impossible. He illustrated the stakes with a stark contrast: "A millionaire can survive losing a home to a wildfire, but for someone who spent a lifetime to own a single house, losing it means losing everything."

US insurers are already pulling out of wildfire-prone areas of California and flood-prone parts of Florida. Garrido also introduced the Actuarial Climate Index, a measure that converts climate risk into an index analogous to the consumer price index, and described how it can be applied to risk measurement and product design.

The final session turned to data innovation. Sheldon Lin, a professor at the University of Toronto, presented techniques for identifying normal, drowsy and aggressive driving patterns using telematics data, while Tim Boonen, a professor at the University of Hong Kong, proposed a pricing model for weather-index crop insurance that incorporates deep learning.

The seminar's overarching message was clear: as AI and climate forces redraw the map of risk, the role of insurance must change with it. For South Korea, the event left an urgent task — reaching swift social consensus if the industry is to move beyond its first steps. "We are still at the stage of just learning to walk," Choi said, "but ultimately we have to start by pooling the capabilities of businesses and individuals alike."


psj@heraldcorp.com