Who Proves What When AI Decides? The Accountability Gap in Global Grain Supply Chain Decisions

AI-assisted procurement, counterparty selection, pricing, and logistics raise a basic question for global grain supply chains: can an enterprise reconstruct who decided what, when, and on what information? This paper distinguishes product provenance from decision provenance and develops a four-layer accountability-void framework covering definition, regulation, enforcement, and precedent. The analysis examines the EU, the United States, China, Brazil, Singapore, Japan, and the WTO/FAO institutional setting, alongside five hypothetical risk scenarios. In the public records reviewed through February 2026, it identifies zero directly matching enforcement actions and judicial or arbitral decisions addressing the allocation of responsibility between AI recommendations and human approval in grain procurement and trading. Existing legal duties provide important points of entry, while the connection between those duties and decision-level evidence remains the central institutional problem. The paper explains how nominal human oversight and fragmented records complicate that connection, and sets out a research agenda for evidence that links data, model output, human review, compliance screening, and execution.