Even Artificial Intelligence Needs a License: When Automation Crosses the Line in U.S. Customs Trade
Artificial Intelligence is transforming virtually every industry, and international trade is no exception. But when automation starts crossing regulatory boundaries, things get a lot more complicated than they might seem at first glance.
In January 2026, CBP (U.S. Customs and Border Protection) quietly published an official ruling that many people in the industry barely noticed, but it significantly changes the game for anyone developing or using AI tools geared toward imports into the United States. The document, identified as CEE.HQ H350722, was published on January 16 and addresses a very straightforward question: how far can an automated platform go before it crosses the line into what is considered customs business, meaning regulated customs activity.
And CBP‘s answer matters a great deal, especially for those building Artificial Intelligence agents for the import market. The ruling did not come out of nowhere. It was prompted by a real case involving a foreign company without a license that operated a digital platform for importers without any licensing from the regulatory agency. What looked like just another automation tool ended up raising serious questions about the limits of what technology can legally do within the American customs process.
If you work in customs compliance, develop AI solutions for international logistics, or simply import products into the U.S., this is a topic you need to follow closely. 👇
The Case Behind Ruling H350722
To understand the weight of this ruling, it helps to dive into the actual case that prompted it. A foreign company, identified in the documents only as Unlicensed Company, was operating an online platform aimed at American importers. This company did not hold a customs broker license issued by CBP and had not sought any kind of approval from the agency to operate in this market.
The platform offered four main services to its users:
- Connecting importers with licensed customs brokers, functioning as a sort of marketplace for import services.
- Automated data extraction from shipping documents using an optical character recognition tool, commonly known as OCR.
- Generating tariff classification suggestions within the HTSUS (Harmonized Tariff Schedule of the United States) using an Artificial Intelligence model.
- Completing and submitting CBP Form 5106 on behalf of new importers.
The central question that CBP needed to answer was straightforward: is this company conducting customs business without the license required by law? And the answer, as we will see, was not a simple yes or no. The agency analyzed each service individually, creating a set of criteria that now serves as a reference for the entire technology market applied to customs trade.
What Is at Stake: Understanding the Concept of Customs Business
To understand the real impact of this ruling, you first need to grasp what CBP means by customs business. Put simply, it refers to any activity that involves transacting business with U.S. Customs on behalf of third parties, such as classifying goods, calculating duties, preparing import entries, or any other procedure that directly affects how a shipment is processed at the border.
These activities, by law, require specific licensing, and anyone performing them without proper authorization is technically operating in violation of American regulations. The problem is that with the arrival of automated platforms, that line has become far blurrier than it used to be.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
In the ruling itself, CBP acknowledged that the definition of customs business is quite broad. This means that AI developers can, even unintentionally, build parameters and features into their agents that exceed what is permitted for unlicensed entities. And that is exactly what makes this ruling so relevant for the tech industry.
The Four Key Takeaways from the CBP Ruling
The CBP analysis was detailed and addressed each of the services offered by the platform separately. Here are the main points of the ruling, which now function as a practical guide for anyone working with automation in customs trade:
1. Connecting importers to brokers: allowed, but with caveats
CBP warned that an unlicensed entity cannot act as an intermediary between a broker and an importer if it is actively participating in decisions and activities related to the preparation or submission of customs documents for imported goods, or any other action that constitutes customs business.
However, since the company in question was not actively participating in decisions about what information should be transmitted to the broker for entry purposes, nor participating in the transmission of those documents and data to CBP, the agency concluded that this specific function did not constitute unauthorized conduct of customs activity. In other words, functioning as a connection point between parties is acceptable, as long as the platform does not make decisions about the content of the information flowing through it.
2. Data extraction via OCR: depends on who decides what goes into the entry
This point is particularly relevant for anyone developing document automation tools. CBP was categorical: an unlicensed entity cannot decide which data should appear in an import entry, regardless of whether that extraction is done manually or through an automated OCR tool.
The agency emphasized that even if the company used an OCR tool developed by another unlicensed entity as part of its platform, CBP would still consider the activity as unauthorized conduct of customs business. The issue is not the technology itself, but the role it plays in the decision-making chain of the customs process. If the tool is effectively selecting and directing data to compose an entry, that constitutes regulated activity.
3. Tariff classification by AI: the six-digit line
This is arguably the most impactful point of the ruling for the Artificial Intelligence ecosystem applied to international trade. CBP determined that an unlicensed entity, even when using AI tools, cannot derive HTSUS subheadings beyond the six-digit level. More detailed classifications require the involvement of a licensed customs broker, especially when the classification information will or could eventually be used for an import entry.
However, if the company’s AI classification tool derives only potential subheadings up to the six-digit level, then the activity is not considered customs business. This distinction between six digits and more than six digits creates a very clear technical boundary for developers of automated classification tools. In practice, this means that language models and tariff classification algorithms need to be designed with a safeguard that prevents suggestions beyond that level, unless they are operating under the supervision of a licensed professional.
4. Submitting CBP Form 5106: reserved exclusively for licensed brokers
The final point was the most straightforward of all. Since the preparation and electronic transmission of documents intended for CBP fall within the legal definition of customs business, only a licensed customs broker can complete and submit CBP Form 5106 on behalf of another party. There is no room for interpretation here. Any platform that automates this process without the involvement of a licensed professional is operating outside the law.
What This Means for the Future of AI Tools in Customs Trade
The identity of the company involved in the case remains confidential, but the implications of the ruling are public and far-reaching. We know that there are several other AI tools currently in development, designed specifically to automate the highly complex world of U.S. imports. With this ruling, many of those platforms will have to rethink their plans significantly.
CBP has effectively set the new order for digital trade: AI can connect humans to humans, but it cannot replace licensed American customs brokers, whether in those professionals’ administrative or analytical capacity. That is a clear message to the tech market.
The market for Artificial Intelligence tools in customs trade has grown rapidly in recent years, driven by the promise of reducing operational costs, minimizing human error, and speeding up processes that historically consumed significant time and specialized labor. Modern platforms are already capable of classifying goods using language models trained on tariff databases, suggesting import regimes, identifying potential documentation inconsistencies, and even predicting which shipments are most likely to undergo inspection. All of this at a speed and scale that no human team can replicate on its own.
The problem is that the more sophisticated the tool becomes, the harder it gets to define where support ends and regulated activity begins. That is exactly where licensing comes in as an element that can no longer be ignored.
AI and Licensing: A Combination the Market Is Still Learning to Navigate
The CBP ruling makes it clear that the fact an activity is performed by an algorithm does not remove it from the scope of existing regulations. In practice, this means that companies developing AI agents for the U.S. import segment need, more than ever, to involve customs law specialists from the earliest stages of product development, not just during the final legal review phase.
The solution architecture, the data it processes, the decisions it automates, and the way it interacts with importers and brokers all need to be designed with compliance as a functional requirement, not as a layer added afterward. This is a fundamental point: regulatory compliance needs to be a design principle, not a patch.
For foreign companies, the situation is even more delicate. The company cited in case H350722 operated without any regulatory connection to CBP, which, beyond the immediate risk of enforcement action, also creates significant exposure for the importers who used the platform. After all, the party accountable to U.S. Customs for the regularity of an import is the importer themselves, not the technology vendor.
This means that using an AI tool without verifying whether it operates within the legal boundaries of American customs trade can create real liabilities for end users, not just for whoever built the product. 🚨
What Changes in Practice for Customs Compliance Professionals
For compliance professionals, the main message of ruling H350722 is quite clear: the regulatory environment is adjusting to the reality of Artificial Intelligence tools, and anyone working in this field needs to stay ahead of that shift.
This involves, among other things:
- Reviewing contracts with technology vendors to ensure they include specific clauses about regulatory responsibility.
- Precisely mapping which steps of the customs process are being performed or assisted by automated systems.
- Rigorously documenting how each AI tool used in the import process was evaluated from a regulatory standpoint.
- Clearly defining who is responsible for overseeing automated decisions within the organization.
- Ensuring that the outputs generated by the platform fall within the boundaries of what is permitted without specific licensing.
This documentation will become increasingly important as agencies like CBP step up their scrutiny of how digital platforms interact with the actual customs flow.
Another important area of focus is internal team training. Many organizations have adopted automation tools without necessarily training their teams to understand the regulatory limits of those solutions, which creates an operational gap that can become a serious problem in the event of an audit or inquiry from CBP.
The New Role of Customs Brokers in the AI Ecosystem
The ruling opens the door to a broader conversation about the role of licensed customs brokers in this new ecosystem. Rather than competing with Artificial Intelligence tools, these professionals can position themselves as the qualified oversight layer that turns an automated platform into a solution that is legally compatible with regulatory requirements.
This combination — technology handling the heavy lifting of data processing while a licensed professional assumes regulatory responsibility — may be exactly the model that the customs trade market will adopt more broadly in the coming years. It is a scenario where AI does not replace the human but amplifies their capacity, while the licensed professional ensures everything stays within legal boundaries.
Especially as more regulatory agencies around the world begin to follow the path CBP is forging with rulings like this one, this hybrid model could become the industry standard. Countries with complex customs systems, such as Brazil, the European Union, and China, are certainly watching closely how the United States is handling this intersection of technology and regulation. 🌐
At the end of the day, the message is simple: Artificial Intelligence is a powerful ally for international trade, but it needs to operate within the rules of the game. And when those rules involve regulated customs activities, even the most sophisticated algorithm in the world needs a license — or at the very least, someone licensed standing right beside it.
