This article examines artificial intelligence (AI)-assisted international commercial mediation through the interface between the EU AI Act and the Singapore Convention on Mediation. It argues that AI systems may be classified as high-risk where they go beyond administrative support and contribute to outcomes capable of producing legal effects through enforceability. Focusing on structural influence, deployer responsibilities, human oversight, transparency, data governance, and ‘shadow AI’, the article shows how failures in AI governance may affect enforceability under the Singapore Convention. It concludes that legitimate AI-assisted mediation requires dual legitimacy: ex-ante governance under the EU AI Act and ex-post scrutiny under the Singapore Convention.
Journal of World Trade