41.3 F
Washington D.C.
Friday, December 1, 2023

PERSPECTIVE: Using Decision Intelligence to Combat Smuggling Networks

ML-based solutions may help with uncovering new and/or evolving parameters that surface new smuggling trends.

Smuggling – the illicit movement of goods into or out of a country – is a major challenge for customs administrations and tax authorities alike. Smuggling plays a key role in the cross-border flow of drugs, fake medicine, weapons, and more, and it’s an insidious threat to fair trade, with both immediate and long-lasting implications for local economies and residents.

Smuggling debilitates the economy when taxes and customs duties aren’t collected on imports and exports – or when they are collected, but at the wrong rate. Smuggling also fosters unfair competition in the commercial marketplace when vendors don’t report (or falsely report) goods or avoid their tax and duty obligations.

In some regions of the world, particularly developing countries, a significant percentage of trade is likely to be unreported, thus helping smugglers thrive in these environments (and in some cases, smuggling becomes a “legitimate” source of income for corrupt officials), often causing severe and lasting economic drag. The rise of e-commerce is amplifying these challenges, with increasing global trade scales, and involvement of small actors in trade.

The dangers of smuggling can be felt on much larger scales as well. Culture heritage is eroded whenever historical/museum artifacts are smuggled out of a country. Smuggled non-indigenous plants and animals introduce far-reaching ecological and agricultural dangers. Smuggling is used to fund and supply terror networks, and propagate the supply of drugs and fake medicine, thereby increasing the burden on public healthcare and policing.


Smuggling can be broadly classified into three main categories: A) organized cases involving syndicates and coordinated activities; B) individual cases where travelers bring in undeclared dutiable goods (e.g. cigarettes, liquor, cell phones etc.); and C) licensing offenses involving importation of meat/poultry, animals/plants, endangered species, etc., without an import license or health certificate required by the law.

Anti-smuggling is commonly practiced in two ways: Risk assessment of the goods entering/leaving the country, to assess if cargo is potentially problematic. This is typically criteria-matching or rule-base methods based on authorities’ best practices and previous experience with different criteria assigned for tax, security or environmental uses cases.

Risk assessment is the leading method used today to help customs officers understand which goods should be inspected, seized, or suspended. According to WCO data (2021), approximately 73 percent of all border seizures were achieved via risk assessment methods.

However, risk assessment on its own isn’t enough, as in many cases the criteria decided by the customs administration does not cover all the potential cases. Therefore, customs administrations also employ random inspections as part of their efforts to detect illicit smuggling. Inspecting cargo and people at random can be an effective approach in some use cases, since it’s clear one cannot cover all the potential cargo that passes through customs. And it can serve to deter potential smugglers.

Alarmingly, the least effective approach to anti-smuggling today is proactive intelligence, accounting for a slim minority of all illicit goods seized globally. Where risk assessment and random inspections can be effective in many cases, illicit goods aren’t detected until the last minute. With proper, proactive intelligence, smuggling can be exposed far earlier, and appropriate resources could be allocated to prevent it in advance. What’s more, seizures achieved via intelligence gathering are often massive in scale, affecting major shipments.

When it comes to tax evasion, participants in a smuggling network might fulfill the requisite reporting obligations, but the data they provide could be suspect. Tax investigators are trained to look for these discrepancies, detecting when imported/exported goods are deliberately undervalued for tax evasion purposes. By monitoring market prices, investigators can determine norm pricing and identify discrepancies. But people cannot detect discrepancies at large scales unassisted, and big data approaches, including Machine Learning (ML)-based mechanisms, must be used.


Many customs administrations and tax authorities rely on BI solutions to combat smuggling, with limited effectiveness. BI queries are static in nature, and based on rules that can quickly become obsolete. BI queries are also developed based strictly on the expertise and experience of the analyst – which is valuable but limited compared to ML-based approaches.

ML-based solutions may help with uncovering new and/or evolving parameters that surface new smuggling trends. While BI is great for many applications, it’s mainly optimized for big-picture analysis and cannot provide targets or surface methods by itself, or address specific use cases that cannot be handled from a bird’s-eye view.

BI solutions suffer from another common drawback in that they typically can only accommodate structured data (databases, spreadsheets, documents, etc.). Hugely valuable unstructured data – images, video, handwritten records – is frequently ignored or must be reviewed manually.

Unstructured data accounts for a fast-growing percentage of today’s available data, and the volume of this data is growing exponentially, sourced from CCTV cameras, open-source databases, and other forums and formats. BI solutions that exclude unstructured data are therefore foundationally limited in their capability and value.


Decision intelligence platforms that provide multifeatured data fusion and ML-based analysis capabilities are naturally beneficial for establishing a fuller investigative picture when combating smuggling and tax evasion. With improved decision intelligence, customs administrations and tax authorities can make better, faster assessments, solving investigations quicker and more effectively, which should act as a deterrent to smugglers and tax evaders and reduce these crimes over time.

A decision intelligence platform offers major advantages. A comprehensive solution can fuse all data sources – structured and unstructured – together in one place, with all the cost and workflow efficiencies that entails. Moreover, data fusion is essential for connecting the dots of an investigative picture, ensuring troves of valuable unstructured data can be leveraged effectively.

Data analytics capabilities utilizing ML are likewise essential for surfacing deeper insights from investigative data. The ability to cross-reference and assess a wealth of data quickly and effectively makes it significantly easier to identify, track and predict criminal patterns.

Lastly, a decision intelligence platform can provide a collaborative environment for internal teams, and internal and external teams working together. A centralized decision intelligence platform also enables the enrichment and visualization of data from multiple angles to provide a much more holistic view of the data and the insights within.

Unlike with BI or tactical risk assessment, decision intelligence enables a fuller, faster investigation to better identify and combat smuggling enablers, networks and beneficiaries, while helping customs administrations and tax authorities adapt to evolving smuggling trends and techniques into the future.


The views expressed here are the writer’s and are not necessarily endorsed by Homeland Security Today, which welcomes a broad range of viewpoints in support of securing our homeland. To submit a piece for consideration, email editor @ hstoday.us.

Tom Saltsberg and Noam Zitzman
Tom Saltsberg and Noam Zitzman
Tom Saltsberg is a Product Manager for NEXYTE, Cognyte's decision intelligence platform, and he has extensive experience in leveraging big data analysis and AI for intelligence investigations. He began his career as an intelligence officer in the Israel Defense Force (IDF) 8200 Intelligence Unit. Tom brings tremendous expertise in conducting data-driven investigations and utilizing state-of-the-art technologies for intelligence analysis. He has dedicated several years to researching the application of machine learning, advanced data analysis techniques and other technologies to enable governmental organizations to extract valuable insights to speed up and resolve their investigations. Tom holds a BSc in Computer Science and MATAR (excellence program) from the Hebrew University in Jerusalem and is currently pursuing a master's degree in computer science at Reichman University. Noam Zitzman is Chief of Intelligence for Cognyte’s Analytics Platform. Noam was previously the head of the Intelligence Methodology Group for Cognyte’s web intelligence platform. Prior to joining Cognyte, Noam led the strategic analysis group in Terrogence, focusing on research into global terrorism, propaganda and recruitment methods. With more than 20 years of operational experience in Israeli intelligence agencies, in the intelligence wing of Israeli Ministry of Foreign Affairs, and in the Israel Defense Force intelligence corps, Noam combines his previous experience with strong academic proficiency. He is also a research associate in the Van Leer Jerusalem Institute, Forum for Regional Thinking, where he regularly publishes articles. His book, “Soundtrack of Revolutions,” was published in 2017.

Related Articles

- Advertisement -

Latest Articles

Verified by MonsterInsights