In Argentina, business associations have called for the reinstatement of value-based criteria or similar systems. The growth of international trade and the digitalization of operations have presented customs administrations with a complex challenge: controlling millions of import transactions in real time while ensuring the accurate determination of customs value. In this context, artificial intelligence (AI) is beginning to play a central role in modern customs risk management systems.
Over the past two decades, the world's most advanced customs agencies—such as those in the European Union, Japan, South Korea, and Singapore—have incorporated big data analytics systems and machine learning algorithms to detect irregularities in import declarations. These systems do not replace the customs valuation process established in the GATT-WTO Valuation Agreement, but they do allow for the highly accurate identification of transactions that present a greater risk of under-invoicing or value manipulation.
Going back to outdated systems, closer to the last century than using modern tools, does not seem like an optimal way of thinking.
The principle remains the same: the transaction value
From a legal point of view, the international customs valuation system continues to be based on the principle of transaction value, that is, the price actually paid or payable for the goods when they are sold for export to the country of import.
This implies that the determination of customs value remains a normative process based on the methods provided for in the Valuation Agreement: transaction value, identical goods, similar goods, deductive value, computed value and last resort method.
However, what has changed radically is the way customs detects suspicious transactions. That's where artificial intelligence has become a strategic tool.
From risk profiles to intelligent analysis engines
Historically, customs used relatively simple risk profiling systems,
based on predefined rules. For example:
- certain countries of origin
- certain products
- certain importers
If an operation matched those parameters, it was selected for control.
Today the paradigm is different. New risk management platforms use big
data and machine learning to analyze millions of historical transactions and detect
anomalous patterns in declared prices.
These systems can automatically compare variables such as:
- price per kilogram or unit
- importer history
- price behavior for the same tariff position
- country of origin and supplier
- characteristics of the commercial operation
When the system detects significant deviations from the expected behavior
Normal statistical standard, generates an automatic risk alert.
international experiences
In the European Union, customs risk management systems integrated within the
Digital customs architecture allows cross-referencing data from millions of declarations to
Detect anomalies in real time. Statistical analysis engines generate alerts.
when a declared price deviates significantly from the usual values for a
specific product.
Japan, for its part, has been operating one of the most advanced electronic systems for decades.
advanced in the world, the NACCS (Nippon Automated Cargo and Port Consolidated
System), which combines electronic processing of declarations with tools for
Predictive analytics to prioritize inspections.
One of the most advanced cases is that of South Korea, whose customs administration
developed the UNI-PASS Smart Customs system, which incorporates artificial intelligence and
Big data analysis to detect commercial fraud, undervaluation, and triangulation
suspicious. I had the opportunity to see firsthand in Ecuador the system they bought from Korea, and the need to update customs computer systems is clear.
Argentina.
Even China has made progress in the use of “customs intelligence” platforms, where
algorithms simultaneously analyze commercial, logistical, and inspection information to
identify inconsistencies in the statements.
What artificial intelligence can—and cannot—do
It is important to clarify a fundamental point: artificial intelligence does not determine the value in
customs.
The determination of value remains a legal process based on regulations
international and in the commercial documentation of the transaction. Artificial intelligence
It fulfills another role: identifying operations that deserve further analysis.
In other words, AI functions as an advanced risk management system, which
It helps customs administrations to concentrate their control resources where
There is indeed a higher probability of irregularities.
This approach responds to the principle increasingly adopted by administrations
modern: intelligent, data-based control, not indiscriminate mass controls.
An opportunity to modernize customs systems
The incorporation of artificial intelligence into value control systems opens up a
This is an important opportunity to modernize customs processes in many countries.
An automated analysis system could, for example:
- detect anomalous prices in real time
- compare operations with international databases
- identify patterns of triangulation or commercial fraud
- generate dynamic risk indicators
This would allow for a reduction in unnecessary controls and a focus on audits of operations.
truly relevant.
Furthermore, an artificial intelligence-based approach could help solve one of the
Recurring problems in many customs systems: excessive discretion in the
valuation processes.
By relying on statistical analysis and objective evidence, intelligent systems can
improve the transparency and predictability of customs control.
The future of customs control
The World Customs Organization has been promoting the concept for several years.
of “Smart Customs”, which combines digitization, artificial intelligence and advanced analytics
data to improve the efficiency of border control.
In that scenario, it is likely that in the coming years we will see the development of engines
global value analysis tools, capable of automatically comparing transactions of
international trade in multiple jurisdictions.
The customs office of the future will not necessarily be a customs office with more controls, but a
customs with better tools to identify where the real risks are.
And in that process, artificial intelligence is emerging as one of the most important tools.
powerful tools to achieve a balance between trade facilitation and effective customs control.
The author is a Specialist in International Trade and holds a Master's degree in Tax Administration and Public Finance, with a solid academic background and extensive experience in foreign trade and customs policies. He teaches at the National University of Córdoba (UNC) and the Catholic University of Córdoba (UCC), where he lectures on courses related to international trade and trade facilitation. He is also an accredited expert of the World Customs Organization (WCO) and a specialist in trade facilitation.
