The Importer of Record Is Becoming a Data Problem

Customs compliance has traditionally been treated as a transaction-level exercise. Classify the product, calculate the duty, file the entry, and retain the records. That model is becoming harder to sustain as customs authorities demand more visibility into the companies, products, and supply chains behind each import.

The shift is especially visible in the United States. Executive Order 14411, “Strengthening Customs Enforcement,” signed June 3, 2026, directs the Department of Homeland Security and US Customs and Border Protection (CBP) to strengthen importer of record requirements, increase vetting, and establish additional disclosure and certification requirements. The order also directs the government to require information or documentation that foreign exporters were required to submit to foreign customs authorities before exporting goods to the United States.

For importers, this creates a challenge that is fundamentally about data. The relevant information may already exist, but it often lives in different systems, under different descriptions, and with different owners. AI can help connect those pieces, but only if companies treat AI as part of a controlled compliance process rather than as a replacement for customs expertise.

Why the IOR Matters More Than It Used To

CBP has already taken a concrete step toward greater scrutiny of importer identity data. In an August 2026 Federal Register notice, the agency said it would begin enhanced enforcement of information submitted through CBP Form 5106. Beginning September 18, 2026, CBP said it could void an importer of record number when the information on file is inaccurate or incomplete, making the number invalid for entering merchandise.

That matters because an IOR record is more than an administrative profile. It connects an entity to customs entries, duties, brokers, and regulatory obligations. If the underlying identity information is wrong, other records can become harder to reconcile.

This is an area where AI and automation can provide practical value without making compliance decisions on their own. Systems can identify conflicting addresses, entity names, contact information, or ownership data across internal records. They can flag records that have not been reviewed recently or identify changes that should trigger human attention. The goal is not to have an algorithm decide whether an importer is compliant, but to make it easier for compliance teams to see where their data may no longer support that conclusion.

Foreign Export Documents Could Expand the Data Burden

The next challenge is potentially much larger. In its September 2026 Advance Notice of Proposed Rulemaking, CBP asked for public input on requirements that would give the agency greater visibility into supply chains and foreign export documentation.

The proposal is still under consideration. CBP is asking whether importers should submit or retain documents that foreign exporters were required to provide to foreign customs authorities before export. The agency specifically raises examples such as export declarations, commercial invoices, packing lists, certificates of origin, export licenses, bills of lading, and air waybills. CBP also asks whether the importer of record should be responsible for retaining and furnishing those documents and whether reasonable care should apply when importers assess their accuracy.

That last question is particularly important. An importer may not create a foreign export declaration. A supplier, distributor, or logistics provider may create it. Yet the importer may still need to explain how the information relates to the US entry.

The resulting task is less about storing PDFs than about comparing information. Does the product description match? Does the quantity make sense? Is the manufacturer the same? Does the country of origin align? Does the declared value require explanation? If two documents differ, can the importer explain why?

AI Is Most Useful When It Connects Evidence

This is where AI can move beyond the most familiar customs use case of classification. Large language models and other AI systems are particularly useful when information is unstructured or distributed across many documents. A product specification might describe an item one way, a supplier invoice might use another name, and a customs entry might contain a shortened description. A certificate of origin can introduce another set of identifiers. A person who knows the business may understand that all four refer to the same product, but finding those relationships manually across thousands of records is difficult.

AI can help surface those connections. It can extract relevant information from documents, normalize product descriptions, identify potential inconsistencies, and prioritize records for review. It can also retrieve supporting customs rulings, regulatory guidance, and prior decisions when a compliance professional is investigating a classification or tariff question.

The World Customs Organization has recognized this broader potential. Its 2025 report on the adoption of AI and machine learning in customs identified data management, governance, risk management, interoperability, cybersecurity, and workforce skills as important considerations for successful adoption. That is a useful reminder that AI in customs is not primarily a chatbot problem, but is a data and governance problem.

The strongest applications will therefore be tied to specific workflows. An AI system might flag a discrepancy between a supplier document and an entry. A compliance professional can then review the underlying records, determine whether the discrepancy is legitimate, and document the reasoning.

Human Review Still Matters

Customs decisions carry financial and legal consequences, which makes explainability especially important. A system that produces an answer without showing the evidence behind it creates a new compliance risk. For this reason, AI used in trade compliance should preserve the information that supports an output. If a system recommends a classification, users should be able to see the product attributes, relevant source material, and reasoning that informed the recommendation. If it identifies a discrepancy, the user should be able to trace the conflicting records.

This approach aligns with the broader principles in the National Institute of Standards and Technology’s AI Risk Management Framework, which emphasizes AI trustworthiness, including accountability, transparency, explainability, and interpretability. NIST also identifies human oversight as an important part of managing AI systems.

For customs teams, that means AI should generally function as an analyst, not an autonomous authority. It can find evidence, compare records, identify changes, and prioritize work. The human expert remains responsible for evaluating the evidence and making the compliance decision.

Build the Data Foundation Before the AI Layer

The temptation with AI is to start with the model. For customs compliance, the better starting point is the data. Importers should ask whether they can connect the information associated with a product across the systems that support the transaction. That can include: 

• Product specifications
• HTS classifications
• Customs rulings
• Supplier and manufacturer information
• Country-of-origin determinations
• Invoices
• Transportation documents
• Regulatory requirements
• Prior entry data

They should also ask whether changes are visible. If a supplier changes a product specification, if a tariff measure changes, or if an importer’s corporate information changes, can the compliance team identify the affected records?

Data quality becomes even more important when AI is introduced. An AI system cannot reliably reconcile records that are incomplete, outdated, or inconsistent. It can identify patterns in bad data, but it cannot turn missing evidence into reliable evidence.

This is consistent with recent work from the WCO, which has emphasized data governance as a foundation for AI and analytics in customs. The organization’s 2026 discussion of AI in customs highlighted the importance of reliable, well-governed data for producing dependable results.

The Next Customs Advantage May Be Explainability

The emerging customs environment is likely to reward companies that can do more than produce an answer. They will need to show how they reached it.

Executive Order 14411 and CBP’s proposed approach to foreign export documentation point toward greater visibility into the parties, products, and records behind an import transaction. The result could be a compliance environment in which disconnected spreadsheets and document repositories become difficult to manage. AI can help, but the most valuable systems will not be those that simply generate classifications or summarize regulations. They will be systems that connect evidence, surface discrepancies, preserve sources, and keep people in the loop.

For importers, preparation can begin now: 

• Validate IOR information
• Map where supplier and foreign export documents are created and stored
• Identify differences between supplier records and US entry data
• Review which compliance decisions still depend on manual searches
• Determine where AI can reduce the work required to find and evaluate the evidence

The future of AI in customs may be less about replacing the customs expert and more about giving that expert a much better view of the transaction. As customs enforcement becomes more data-driven, the competitive advantage may belong to the importers that can answer a simple question with confidence: not just “What did we declare?” but “Can we show why it was correct?”