"AI-Related Damages Require Presumption of Causation"

Seeking the Future Direction of Civil Law in the Age of AI
Insights from the Henri Capitant Society World Congress in Seoul

As countries around the world are responding to civil liability issues related to artificial intelligence (AI) by adjusting their existing legal systems, legal scholars gathered in Seoul have agreed that structural barriers such as proving negligence or causation are major obstacles to remedying victims, and that practical solutions to these challenges must be sought.

Opening ceremony of the '2025 Henri Capitan Society World Congress' held on the 16th at Seoul National University Gwanak Campus. Photo by Legal Times

Ko Yugang, a professor at Seoul National University School of Law, pointed out at the '2025 Henri Capitan Society World Congress (Journees Internationales Henri Capitant 2025)' held on June 16 that "most countries are addressing risks posed by AI through their existing legal frameworks, and very few have introduced entirely new legal systems." In his presentation on "Artificial Intelligence and Civil Liability," Professor Ko predicted that "recent developments such as the European Union's AI Act will have a significant impact on future legislative trends." While the EU's AI Act does not directly stipulate civil liability, it imposes requirements such as risk-based classification, transparency, explainability, and documentation obligations. As a result, failure to comply with these requirements is likely to be considered negligence in future civil liability determinations.

The discussion also continued on the question of "who is at fault when damage occurs due to an AI decision." Professor Ko analyzed that "so far, AI errors have typically been attributed to human fault, such as design flaws, biased data, or insufficient supervision," and explained that "compliance with the duty of care is a crucial standard for judgment, and this standard will be further defined by the AI Act going forward."

It was also anticipated that "proving causation," which determines whether the negligence was the direct cause of the damage, would become a major issue. This is because it is not easy to clearly identify, for example, why an AI financial advisory algorithm made a wrong decision. According to legal scholars worldwide, AI faces barriers to proof because: ▲ the process by which it produces results is opaque, ▲ it learns autonomously without direct human commands, and ▲ it is interconnected with various systems, making proof virtually impossible. They pointed out that these three elements?AI's opacity, autonomy, and complexity?fundamentally challenge the core of civil liability law.

The burden of proving negligence and causation is becoming a factor that threatens the effectiveness of victim protection. This means that even if someone suffers harm due to AI, there may be cases where they are unable to receive compensation. In response, legal scholars proposed shifting the burden of proof to the business operator by presuming negligence or causation under certain conditions. This would legally presume causation if specific criteria are met. They also mentioned requiring businesses to submit technical information through disclosure orders to ensure victims have access to relevant technology. However, they emphasized the need to establish fair standards regarding the scope of exoneration and the final allocation of responsibility among multiple liable parties.

During the discussion following Professor Ko's comprehensive presentation, delegations from countries including Colombia, Chile, Japan, Italy, and France shared a variety of perspectives.

On June 17, Razvan Dinca, Dean of the Faculty of Law at the University of Bucharest, led a discussion on "Artificial Intelligence and Copyright." He explained, "The interests and issues related to the protection systems applied to AI, data, and their outputs are complex and sometimes conflicting," and added, "Countries around the world are emphasizing the need to clarify the attribution of rights to generated content and to strike a balance between protection, innovation, freedom, and responsibility." That afternoon, nine topics were discussed in depth during breakout sessions (ateliers), including: △ AI and legaltech and applications △ AI and the judiciary △ Criminal offenses by AI △ AI and discrimination △ Predictive policing and fundamental rights △ Korea's Basic AI Act and economic effects △ Cybersecurity and AI △ AI in industry △ Preventive measures related to AI.

This conference, co-hosted by the Korean-French Civil Law Association (President Lee Eunhee) and the Association Henri Capitant, was held from June 16 to 18 at the Centennial Memorial Hall for Modern Legal Education at Seoul National University Gwanak Campus in Sillim-dong, Gwanak-gu, Seoul. The Association Henri Capitant, an international legal society founded in 1935, holds a world academic congress every June and is celebrating its 90th anniversary this year. Korea joined the society in 2012, with the Korean-French Civil Law Association serving as the Korean branch.

Approximately 160 participants from about 25 countries attended this year's congress, including around 140 international participants. The opening ceremony was attended by Lee Eunhee, President of the Korean-French Civil Law Association; Philippe Dupichot, President of the Association Henri Capitant; Philippe Bertoux, French Ambassador to Korea; Supreme Court Justice Shin Sookhee (56, Judicial Research and Training Institute Class 25); Ko Haksoo, Chairperson of the Personal Information Protection Commission; and the keynote speech was delivered by Ha Jungwoo, Head of the AI Center at NAVER Cloud.

Supreme Court Justice Shin Sookhee stated, "The law should not merely follow technological progress but should play a role in anticipating and preparing for potential legal issues in the future," and added, "This event will serve as an opportunity to explore the future direction of our civil law in a technological environment." Chairperson Ko Haksoo noted, "As artificial intelligence expands into the realms of human judgment, creativity, and decision-making, it poses significant challenges to the existing legal order and personal information protection," and emphasized, "AI and personal information protection are closely related."

Seo Hayoun, Legal Times Reporter

※This article is based on content supplied by Law Times.

This content was produced with the assistance of AI translation services.