Over‑ and Under‑Invoicing: The TBML Risk Hiding in Plain Sight
Over‑ and under‑invoicing is one of the most persistent—and hardest to detect—forms of Trade‑Based Money Laundering (TBML). The challenge isn’t a lack of rules. It’s the difficulty of knowing what goods are really worth.
In global trade, legitimate prices vary widely based on quality, brand, specifications, market conditions, trade routes, and timing. Traditional controls, built largely around Harmonized System (HS) codes, simply can’t capture this complexity. Broad averages mask genuine variation, while giving criminals plenty of room to manipulate prices without raising suspicion.
Why Price‑Based TBML Is So Hard to Detect
HS codes group vastly different goods under a single classification. A category such as “men’s cotton shirts” may include everything from low‑cost bulk garments to premium branded products, yet pricing benchmarks treat them the same.
Criminals exploit these gaps by setting prices that sit comfortably within “acceptable” ranges, even when the transaction is deliberately mispriced. As a result, financial institutions struggle to distinguish legitimate commercial variability from intentional value transfer.
Regulatory Pressure Is Rising
Regulators are paying close attention. FATF and industry bodies continue to highlight pricing manipulation as a critical TBML risk requiring stronger controls.
The scale is significant:
The message is clear: price‑based TBML detection can no longer rely on rough estimates and manual judgement alone.
The Limits of Traditional Price Checks
Most institutions still depend on fragmented approaches, external HS price ranges, manual web searches, or analyst‑led comparisons with historical transactions. These methods are slow, subjective, hard to scale, and inconsistent across teams and regions.
They also struggle with unstructured data. Goods descriptions vary widely, HS codes are often missing or incorrect, and subtle differences in terminology can completely change the meaning and value of what’s being traded.
How AI Changes the Equation
AI and Large Language Models (LLMs) make it possible to assess pricing risk with far greater precision. By analyzing detailed, unstructured trade data such as product descriptions, origin and destination, timing, brand, specifications, and real‑time market signals: AI can infer realistic, context‑aware price ranges that reflect how goods are actually traded.
This shifts TBML controls from static rules to intelligence‑driven assessment, designed to detect schemes that deliberately mimic normal market behavior.
Eastnets SafeTrade: AI‑Powered TBML Detection
Eastnets SafeTrade embeds AI‑driven price intelligence directly into its Trade‑Based Financial Crime (TBFC) controls. Delivered as a bundled, white‑labelled service, it removes the need for customers to engage directly with external AI providers, reducing complexity, risk, and operational overhead.
SafeTrade’s AI engine:
All AI processing is fully managed within the SafeTrade platform, ensuring secure, scalable, and consistent price‑based risk detection.
From Rules to Intelligence
As TBML schemes grow more sophisticated, checklist‑driven controls are no longer enough. AI enables financial institutions to move from reactive monitoring to proactive, contextual intelligence, strengthening compliance, reducing false positives, and staying ahead of evolving regulatory expectations.
With SafeTrade, AI becomes a practical tool for protecting the integrity of global trade.