Artificial Intelligence in International Arbitration

Abstract

Generative artificial intelligence (AI) is now being used in international arbitration for tasks such as legal research, document review, translation, drafting, evidence analysis, and case administration. The main legal instruments of international arbitration, including the 1958 New York Convention, the UNCITRAL Model Law, and the major institutional rules, were written before generative AI became widely available. They therefore do not contain detailed AI provisions. They do, however, contain rules dealing with equal treatment, the opportunity to present a case, the authority and responsibility of the tribunal, reasoned awards, confidentiality, and enforcement. This article examines how those existing rules can apply when AI is used by arbitrators, counsel, or parties. It uses a doctrinal and comparative approach, looking at primary arbitration instruments, institutional AI guidance, selected national judicial guidance, and recent decisions involving AI-generated legal authorities. The main argument is that the important distinction is not simply between using AI and not using AI. The more useful distinction is between AI used as an assisting tool with proper human checking and AI use that has a real effect on the tribunal’s judgment, procedural fairness, confidentiality, or the reliability of the case record. The article also develops a risk-based framework based on four questions: who remains responsible for the decision; how important the AI use was to the proceedings; whether the affected party could understand and respond to the AI-derived material; and whether the problem creates a legally recognised defect under the law of the seat or the enforcement forum. The article concludes that existing law can deal with many AI-related problems without an immediate amendment to the main international instruments, but institutional rules and procedural orders should give clearer guidance on human control, material disclosure, verification, confidentiality, and record-keeping.