Inside Chinese Money Laundering Networks and how Federated AI can detect and prevent them
Written by: Board Director of Consilient and CEO of IDPL Consulting, LLC and Matthew Girgenti, Managing Consultant at Jetty Partners, LLC
Despite spending billions of dollars enhancing compliance programs and deploying increasingly sophisticated technologies and AI-powered algorithms, financial institutions have struggled to prevent and detect the hundreds of billions of dollars of Chinese-laundered money. These activities enable the activities of Mexican drug cartels, the importation of fentanyl, the promotion of human trafficking, fraud, and various other criminal activities.
At the same time, such money laundering activities have been linked to a variety of commercial activities, including, among others, the sale of electronics and luxury goods, health care fraud, senior day care facilities, and real estate purchases.
Chinese money-laundering networks (CMLNs) and Mexican criminal organizations have developed a large, agile, and increasingly sophisticated ecosystem aimed at evading legal authorities in China, Mexico, and the United States. A recent “Financial Trend Analysis” from the U.S. Department of the Treasury’s Financial Crimes Enforcement Network (FinCEN) noted that the agency reviewed more than 137,000 Bank Secrecy Act filings spanning the five-year period from 2020 through 2024 and identified more $300 billion in suspicious activity during that time. FinCEN concurrently issued an advisory urging financial institutions to remain vigilant in identifying transactions potentially related to these operations.
The cost of failure is high. In 2024, one institution paid more than $3 billion and pleaded guilty to criminal charges after failing to prevent laundering activity, including a single customer who moved $470 million in drug proceeds through the bank. To mitigate such risks, FinCEN and other regulators encourage depository institutions to consider, evaluate, and, where appropriate, responsibly implement innovative approaches to meet their AML compliance obligations.
Among these, federated AI stands out as a powerful yet underutilized tool that allows institutions to share intelligence and detect suspicious activity, without exposing sensitive customer or competitive data.
What are Chinese Money Laundering Networks (CMLNs)?
Mexico and China both place restrictions on the flow of US currency into their country. Since 2010, Mexican law has limited the amount of USD that customers can deposit into domestic financial institutions. Meanwhile, China places limits on the amount of Chinese currency its citizens can convert into U.S. dollars. As a result, Mexican criminal organizations cannot easily move the proceeds of their activities into local currency, while Chinese nationals have difficulty obtaining large quantities of U.S. dollars.
Enter CMLNs.
According to FinCEN, CMLNs are decentralized groups that combine informal value transfer techniques, trade-based schemes, complicit merchants, and recruited “money mules” to convert bulk cash into legitimate-looking assets and cross-border value. CMLNs use a variety of tactics and operate in multiple countries, but the basic operation of CMLNs involves a trust-based system that employs simultaneous or near-simultaneous “mirror” transactions in each of the countries whose currency they are laundering.
In one simple example, a CMLN receives USD from the cartel’s US-based actors and transfers an equivalent value of pesos into the cartel’s Mexican accounts. The US-based CMLN counterparts then sell the USD they “purchased” to Chinese buyers who transfer the equivalent amount of Chinese currency into a China-based bank account. The Chinese buyers then use “mules” to deposit the US currency into US financial institutions for use by their Chinese clients.
This money is often further laundered through the purchase of cashier’s checks, real estate, or luxury goods. The CMLN profits by charging fees to their Chinese and Mexican clients to facilitate these transactions. As noted above, in addition to the drug trade, authorities have linked CMLNs to a wide range of criminal enterprises, including human trafficking, fraud, and illicit gaming. CLMNs operate globally, and may coordinate with professional money launderers in other countries, such as Colombia.
Red flags for detecting Chinese Money Laundering Networks
Financial institutions must file a Suspicious Activity Report (SAR) if they know, suspect, or have reason to suspect that a transaction is connected to illegal activity. As part of its advisory, FinCEN identified 18 red flags to help financial institutions detect transactions involving CMLNs.
While no single indicator is determinative, these red flags include:
- ➡Individuals making large cash deposits inconsistent with their reported occupation.
- ➡Same-day funding of wire transfers to foreign accounts.
- ➡Use of Chinese passports with suspicious documentation.
- ➡Large purchases of cashier’s checks for real-estate closings.
- ➡Rapid inflows and outflows from newly opened accounts.
- ➡Unusually high volumes of credit-card purchases of high-value goods.
In addition to the routine reporting of cash transactions exceeding $10,000, federal law requires banks and other financial institutions to conduct appropriate due diligence with respect to their customers. As set out by FinCEN, these due diligence programs “must include risk-based policies, procedures, and controls designed to identify and minimize risks associated with foreign agents and counterparties.” To facilitate this, FinCEN encourages information-sharing among institutions as a tool to identify such individuals and transactions.
How Federated AI strengthens AML detection and collaboration
Federated AI enables financial institutions to collaborate on detecting financial crime, such as money laundering and fraud, without directly sharing sensitive customer data. Using a federated learning platform, each participating institution trains a model on its own data behind a secure firewall.
The model’s learned parameters—but none of the underlying customer data—are then transmitted to a central server or coordinating hub. From there, the hub aggregates these parameters into a global model, which is redistributed to all participants. While privacy laws, data-localization rules, and competitive concerns may prevent pooling of transaction-level data, federated learning allows machine-learning models to train across multiple institutions while keeping data local.
Enhancing AML detection capabilities
For anti-money laundering (AML) applications, Federated AI makes it possible to identify patterns that no single institution could detect in isolation, such as mule-account networks or cross-bank activity. By learning from patterns across multiple datasets, the collective model becomes more accurate at identifying anomalous transaction behavior and methodologies consistent with CMLN activity.
Building collective defense
With Federated AI, each participating institution benefits from broader risk insights and stronger detection models without breaching confidentiality or competitive boundaries. This architecture represents a promising path to achieving both data protection and collective defense.
As money launderers exploit fragmented oversight, federated learning offers a way to coordinate privacy-preserving financial crime prevention across institutions. Financial institutions that implement Federated AI and other advanced technologies into their AML programs not only reduce the risk of being an unwitting party to illegal transactions but can also demonstrate to regulators that they are using every available tool to meet their due-diligence obligations.
Learn more about how federated AI can strengthen your AML strategy:
