An exclusive article by Fred Kahn

Trade-based money laundering (TBML) remains one of the most elusive challenges for compliance teams worldwide, even as the digital arms race in financial crime detection continues to accelerate. Financial institutions, regulators, and technology providers often tout artificial intelligence (AI) and blockchain as game changers in this space. But has technology truly made it easier for compliance professionals to detect red flags, or does reality lag behind the promise?

AI and Blockchain in TBML Detection: Tools, Pilots, and Practical Use

For years, compliance professionals have recognized that TBML presents a uniquely difficult set of challenges. Traditional anti-money laundering controls, including rules-based transaction monitoring and KYC checks, often fall short in complex international trade environments. Criminal organizations exploit the opacity, scale, and fragmented nature of global trade—using tactics like over-invoicing, under-invoicing, multiple invoicing, and falsely described goods—to move illicit funds.

AI and blockchain, hailed as disruptive technologies, have entered the TBML detection narrative as potential solutions to overcome the limitations of legacy systems. The European Union’s Supranational Risk Assessment, FATF’s 2023 guidance on trade-based money laundering, and industry whitepapers by SWIFT and the Wolfsberg Group all point to data sharing, advanced analytics, and digital ledgers as future pillars of TBML controls.

AI in TBML Detection: Smart but Not Omniscient

AI’s application to TBML detection takes several forms. The most common approaches include machine learning models trained to identify suspicious trade patterns, anomaly detection engines, and natural language processing (NLP) tools that extract and cross-reference trade data from diverse documents.

Pilot programs are multiplying, especially at larger banks and global logistics providers. For example, Citi, HSBC, and Standard Chartered have launched AI-driven trade surveillance pilots in partnership with technology vendors like Quantexa and Silent Eight. These systems scan shipping documents, invoices, customs declarations, and payment flows, seeking inconsistencies or risk indicators that humans might miss.

Despite the promise, industry studies suggest adoption is not yet widespread. According to a 2024 survey by the International Chamber of Commerce, less than 25% of major global banks had fully operational AI-driven TBML systems, while most relied on pilot deployments or rules-based enhancements. AML experts emphasize that data quality and coverage are critical barriers. Trade data is notoriously fragmented, often missing key details about goods, routes, beneficial owners, and counterparties. As a result, AI models trained on incomplete or siloed data risk producing false positives or missing nuanced red flags.

Dr. Justine Walker, Head of Global Sanctions and Risk at ACAMS, noted at a recent FATF event: “AI has potential, but without robust, standardized, and timely data feeds, it’s like using a magnifying glass on a foggy window. We need much better data foundations before we can claim real breakthroughs in TBML detection.”

Blockchain for TBML: Transparency and the Limits of Distributed Ledgers

Blockchain is frequently discussed as a potential game changer for trade finance and TBML risk management. The core idea is that distributed ledgers can provide an immutable, transparent, and shared record of trade transactions, reducing opportunities for document manipulation or double invoicing.

Several consortia—including Marco Polo, we.trade, and Komgo—have piloted blockchain-based trade finance platforms. These systems allow banks and corporates to record letters of credit, bills of lading, and payments on a shared ledger, making it theoretically easier to trace goods, ownership, and money flows.

However, a closer look reveals that adoption is slow, and the hype sometimes outpaces the practical benefits. According to the World Economic Forum’s 2024 survey on blockchain in trade, fewer than 12% of surveyed trade finance transactions were processed on blockchain networks, and many projects were still at the proof-of-concept or limited deployment stage.

One major issue is interoperability. Global trade involves hundreds of stakeholders—exporters, importers, carriers, customs brokers, banks—each using different IT systems, legal standards, and data formats. Blockchain-based platforms are often siloed, unable to communicate with each other or with legacy systems, limiting their ability to provide a single, comprehensive view of trade transactions.

Dr. Shariq Mirza, CEO of Supply Chain Risk Management Inc., commented, “The real promise of blockchain in TBML is end-to-end transparency, but we’re not there yet. Fragmented adoption and lack of standardized data formats mean that most compliance teams still need to piece together information from multiple sources, both on and off-chain.”

Case Studies: Where Technology Succeeds and Fails

Some recent case studies illustrate both the possibilities and pitfalls of using AI and blockchain in TBML detection.

In 2023, a major European bank piloted an AI-based solution to screen cross-border trade flows for red flags. The model successfully flagged a series of suspicious shipments involving undervalued goods routed through high-risk jurisdictions. However, post-mortem analysis revealed that the system generated a large number of false positives—mostly due to inconsistent data fields and lack of context on legitimate trade deviations. Human investigators spent considerable time wading through alerts to identify genuine cases.

By contrast, Singapore’s Infocomm Media Development Authority and several trade finance banks have piloted blockchain-based e-invoicing networks since 2022. These systems did reduce double invoicing and invoice fraud by offering real-time, immutable records accessible to all parties. However, they faced legal and commercial hurdles: cross-border legal recognition of digital documents is still evolving, and not all counterparties are willing or able to onboard to blockchain platforms.

A 2024 study by the Basel Institute on Governance concluded that while both AI and blockchain have produced incremental improvements in certain high-volume, high-risk corridors, neither has yet delivered the “silver bullet” for TBML. Instead, their value lies in enhancing, rather than replacing, traditional AML controls and human judgment.

Practical Limitations: What’s Holding Technology Back?

Even the most advanced AI and blockchain solutions face fundamental challenges when deployed against TBML risks.

• Data Quality and Integration: Trade data is often incomplete, inconsistent, or unstandardized. Many shipping documents are still paper-based, and integrating structured and unstructured data from multiple jurisdictions is a massive undertaking.
• Alert Fatigue and False Positives: Machine learning models are only as good as the data and rules they’re built on. Poorly trained systems often overwhelm compliance teams with excessive alerts, which can erode trust in automation.
• Regulatory Uncertainty: Legal frameworks for both AI and blockchain in AML are evolving. The EU’s Artificial Intelligence Act, FATF’s risk-based approach guidance, and multiple national initiatives are beginning to clarify expectations, but differences in cross-border standards create compliance uncertainty.
• Cost and Complexity: Implementing AI and blockchain at scale requires significant investment in technology, skills, and change management. Many smaller and mid-sized banks lack the resources to adopt these tools beyond limited pilots.
• Human Judgment Remains Central: As of 2025, there are no regulatory regimes mandating or even recommending AI or blockchain as a standalone solution for TBML. The Wolfsberg Group’s 2024 guidance, for example, reinforces the importance of combining technology with expert human analysis and strong governance.

A recent ACAMS and Dow Jones survey highlighted that 64% of AML professionals viewed AI as “useful but not transformative” in TBML, while only 18% felt their organization had seen a measurable drop in undetected TBML risk due to new technology.

Conclusion: A Measured Path Forward for AI and Blockchain in TBML

Despite considerable hype, AI and blockchain remain tools—not panaceas—for TBML detection. They have improved certain aspects of data analysis, alert generation, and trade transparency, particularly in high-volume trade corridors and among larger financial institutions able to invest in pilots and integrations. However, the reality is that practical, regulatory, and operational barriers continue to limit widespread adoption and measurable impact.

Future progress likely hinges on three critical factors:

• Improved data standards and cross-border information sharing, including digitalization of trade documents.
• Regulatory convergence around the use of AI and blockchain in AML controls.
• Ongoing training for compliance teams to interpret and leverage new technology effectively, rather than seeing it as a black box.

For now, compliance teams should approach AI and blockchain as powerful enhancers for existing AML frameworks—useful for surfacing new red flags, but not replacements for strong risk management, rigorous controls, and skilled investigators.

---

Related Links

• FATF: Trade-Based Money Laundering Guidance
• EU Artificial Intelligence Act – Official Document
• Wolfsberg Group Guidance on Trade Finance Principles
• ICC UK launches updated digital trade roadmap

Other FinCrime Central Articles About TBML

• Why Good KYC Alone Can’t Stop Trade-Based Money Laundering
• Complex Trade Documents: A $2 Trillion Gateway Fueling Money Laundering Risks in Trade Finance
• How Free Trade Zones Make It Easier for Trade-Based Money Laundering

Some of FinCrime Central’s articles may have been enriched or edited with the help of AI tools. It may contain unintentional errors.