Vietnam's legal system treats AI harm like traditional product defects, but the technology creates entirely new problems that existing laws cannot adequately address. When an AI system causes damage, it's often unclear who bears responsibility: the developer who built the model, the company that deployed it, the operator who supervised it, or the user who relied on its output. Vietnam's 2015 Civil Code was written long before artificial intelligence became widespread, and it lacks the tools to handle these distributed chains of responsibility.

The challenge goes deeper than just identifying who is at fault. AI systems operate through machine learning processes that are opaque, probabilistic, and only partially controllable by any single actor. This "black box" nature makes it nearly impossible for injured parties to prove how a harmful outcome was produced, let alone establish the causal link required under traditional tort law. Vietnam's current legal framework has no mechanisms to help victims access the technical information they need to build a case.

How Is the EU Solving This Problem?

The European Union has emerged as the most developed regulatory laboratory for addressing AI-related harm, and its approach offers valuable lessons for Vietnam. Rather than relying on a single legal instrument, the EU has constructed a layered framework that combines preventive rules with compensatory remedies.

• Risk-Based Classification: The EU AI Act establishes a framework that categorizes AI systems by the level of risk they pose and imposes corresponding duties on providers and deployers of high-risk systems.
• Evidentiary Support: The proposed AI Liability Directive seeks to reduce the information imbalance between injured parties and AI companies by facilitating access to evidence and introducing rebuttable presumptions of causality under certain conditions.
• Product Liability Expansion: The Revised Product Liability Directive expands the definition of "product" to include software and AI-enabled systems, allowing claimants to seek compensation under a strict liability framework when defective digital products cause damage.

This multi-layered approach represents more than a technical adjustment of existing rules. It signals a broader shift toward a victim-oriented and risk-sensitive model of civil liability, one that acknowledges the evidentiary and structural difficulties inherent in AI litigation.

What Specific Gaps Does Vietnam Need to Address?

Vietnam's legal system faces three critical regulatory gaps when it comes to AI-induced harm. First, Vietnamese law does not clearly define how liability should be allocated among developers, deployers, operators, owners, and users when damage results from autonomous or semi-autonomous outputs. Second, evidentiary mechanisms remain underdeveloped, particularly in cases where claimants need access to technical information to establish fault, defect, or causation. Third, the legal system has yet to adopt a coherent, risk-based approach that distinguishes between low-risk and high-risk AI applications and calibrates liability accordingly.

Although recent legislative developments demonstrate growing awareness of digital risks, they do not yet amount to a dedicated regime for civil liability arising from AI-caused harm. This leaves victims of AI-related injuries in a precarious position, with no clear path to compensation and no incentive for companies to implement stronger safety measures.

What Should Vietnam's Reform Strategy Look Like?

Researchers examining this issue have identified a roadmap for Vietnam's legal modernization. The framework should pursue incremental reforms focused on several key areas: risk differentiation that distinguishes between low-risk and high-risk AI applications, digital product liability that explicitly covers software and AI systems, stronger operator duties that clarify responsibilities throughout the AI value chain, and improved access to evidence mechanisms that help injured parties understand how harm occurred.

Importantly, these reforms should not exist in isolation. They must be embedded within a coherent AI governance framework that links preventive obligations to compensatory remedies. This approach mirrors the EU's strategy and reflects a broader recognition that civil liability alone cannot manage AI risks; it must work in concert with regulatory oversight and industry standards.

The stakes are significant. As AI systems increasingly perform predictive, classificatory, and quasi-decisional functions across public administration, commerce, healthcare, finance, and transportation, the potential for harm grows exponentially. Without clear legal frameworks, victims will be denied redress, companies will lack incentives to implement safety measures, and innovation may proceed without adequate accountability.

Vietnam's legal system stands at a crossroads. By learning from the EU's experience and adapting those lessons to its own institutional context, Vietnam can build a civil liability regime that protects victims, encourages responsible innovation, and ensures that the benefits of AI are not purchased at the expense of legal accountability.