UK Supreme Court's IEEPA ruling and the US CAPE portal launch on 20 April expose how AI-generated tariff classification errors create systemic legal liability for importers and their customs advisers

The US Supreme Court has ruled that IEEPA (the International Emergency Economic Powers Act — the US statute used by President Trump to impose broad tariff packages in 2025) tariffs are unlawful, triggering a refund mechanism for importers who overpaid. The CAPE portal (Customs Automated Processing Engine — the US Customs and Border Protection system for filing refund claims) launches on 20 April 2026 to process those claims. The legal dimension that is directly relevant to AI in law concerns how importers and their legal advisers used AI-assisted tariff classification tools during the period when the unlawful tariffs were applied. A significant number of importers relied on generative AI tools — embedded in customs management software or used directly by in-house counsel — to classify goods and calculate duty liability under the rapidly changing IEEPA tariff schedules. Where those AI tools misclassified goods, importers may have either overpaid (generating a refund claim) or underpaid (generating a potential liability), and the question of legal responsibility for AI-generated classification errors is now live. For UK and EU firms advising importers on CAPE portal refund submissions, the practical challenge is that many clients do not have complete audit trails of the classification methodology used — particularly where AI tools were applied without human review of each classification decision. The EU AI Act, which entered its early application phases in 2025 and 2026, classifies AI systems used in customs and trade compliance as potentially high-risk systems requiring human oversight and documentation obligations. The IEEPA refund wave is therefore a real-world stress test for whether firms and their clients maintained the AI governance documentation that regulators will increasingly require.