Enterprise Liability in Product Liability in the Era of AI: The Study of the New Product Liability Directive of the European Union

Abstract

Artificial intelligence (AI) represents a novel innovation on a global scale, providing significant convenience across various aspects of human life. However, the unique characteristics of AI—its complexity, opacity, and autonomy from human control—fundamentally distinguish it from traditional products. These distinct features inevitably lead to new and complex legal challenges regarding liability, diverging significantly from the conventional frameworks established under existing legal systems. Consequently, it is essential to reconsider and reform the legal framework governing liability for defective products and services in light of these technological advancements.

            This article focuses on analysing the theory of enterprise liability in relation to defective products, particularly those involving AI. It examines how enterprise liability might be applied to AI-related harms. The legal framework of the European Union Directive 2024/2853 on product liability serves as the primary case study, investigating the extent to which the EU Directive aligns with the principles of enterprise liability and how theoretical foundations have been utilised to inform recent reforms.

            The analysis also examines the EU’s recent reforms aimed at addressing AI-specific challenges, including the redefinition of the term “economic operator,” the expansion of the definition of “product” to encompass both tangible and intangible items, and the extension of liability rules to certain services. Furthermore, the article considers presumptions regarding the burden of proof and product defectiveness, which are intended to alleviate evidential difficulties for claimants. These reforms are assessed in terms of both their strengths and limitations.

            Subsequently, the article considers whether Thailand should adopt elements of the EU’s approach to revise its Liability for Damages Arising from Unsafe Products Act B.E. 2551 (2008). The findings suggest that Thailand’s legal framework on liability for defective products should be expanded in several key areas. For example, the definition of “economic operator” should be broadened to include businesses involved in the development and provision of AI, the definition of “product” should be extended to cover both movable and immovable items, whether tangible or intangible, and the scope of the law should be broadened to encompass services alongside products. Additionally, Thai law should introduce legal presumptions regarding product defectiveness and shift the burden of proof to the operator in cases involving AI-related products and services. The definition of “damage” should also be expanded to cover non-material harms, such as loss or destruction of data.

            Crucially, the Thai legal system must also consider including a “development risks defence” for manufacturers, which would provide an exemption from liability where, at the time of placing the product or service on the market, the risks could not reasonably have been known based on the existing scientific or technical knowledge.

            These recommendations aim to align Thai law with the principles of enterprise liability and strike a fair balance between the interests of producers and consumers in the era of artificial intelligence.