The Mexico–U.S. corridor: From fragmented intelligence to risk mitigation

Cross-border financial crime between the United States and Mexico sits at the center of one of the most active financial corridors in the world.

More than $51 billion moves annually from the U.S. to Mexico in remittances alone, creating scale, speed, and constant movement through the system. On top of this, U.S. authorities continue to highlight how transnational criminal organizations move and layer funds across both jurisdictions through coordinated laundering activity.

The result is a system where criminal activity is organized, persistent, and financially significant.

Detection frameworks remain anchored at the institutional level. Monitoring works within those boundaries. The activity does not. Intelligence moves more slowly than the networks it is meant to expose, and often without the context needed to act early.

So instead, institutions are left working with partial visibility. For banks exposed to the Mexico–U.S. corridor, the challenge is connecting activity that spans jurisdictions, counterparties, and time. Relevant signals rarely sit in one place.

Cross-border insight connects those elements. It allows institutions to see activity as it unfolds and strengthens the ability to identify and mitigate risk with greater precision.

Why the Mexico–U.S. corridor is in focus now

The Mexico–U.S. corridor carries high volume and elevated risk.

In the third quarter of 2025 alone, $15.6 billion moved from the U.S. to Mexico through formal remittance channels. At the same time, organized crime linked to cartel activity continues to generate multi-billion-dollar economic impact across the region.

It is exposed to trafficking networks, fraud ecosystems, and organized financial crime, and that activity spans multiple institutions and jurisdictions as part of connected systems.

So the corridor becomes a natural point of focus.

Equally, it presents an opportunity. With the right conditions in place, it can serve as a reference model for how cross-border intelligence sharing operates at scale.

But there’s a challenge banks struggle to overcome.

Criminals are cross-border by design. Intelligence is not

Financial crime in the Mexico–U.S. corridor operates as a connected system. Networks move activity across jurisdictions in real time, with funds, actors, and typologies moving between institutions in coordinated flows.

The challenge is widely recognized. Yet it persists because the system is built around institutional boundaries.

Most frameworks are anchored within institutional and domestic limits. AML controls follow legal entity, regulatory perimeter, and internal data. Those constraints define how risk is assessed and how decisions are made, and they shape what can be seen.

And the outcome is consistent.

Risk is assessed within a single institution. Signals are evaluated against internal histories, internal relationships, and internal thresholds. As a result, activity that spans jurisdictions carries less weight when viewed in isolation.

That becomes visible in how coordination appears. It rarely surfaces clearly at the point of detection. Instead, it is distributed across institutions and only becomes visible once activity is pieced together, often after funds have moved.

Exposure builds at that point.

Network-level activity does not need to present as high-risk within any one institution to move across the system. It only needs to remain unremarkable in each place it touches.

The system is working as designed. The problem is what it was designed to see.

Why cross-border intelligence sharing remains limited

Efforts to strengthen cross-border intelligence sharing have been underway for years, yet progress remains uneven.

The starting point is clarity. Or more accurately, the lack of it.

#1. Legal clarity is inconsistent

Frameworks define the boundaries of sharing, but those boundaries are interpreted differently across jurisdictions and institutions. What can be shared, with whom, and under what protections is not always clear.

As a result, that uncertainty shapes behavior.

Where guidance lacks precision, institutions default to caution. Over time, that caution becomes embedded, and cross-border sharing is shaped more by perceived risk than by operational need.

Ambiguity is the enemy of collaboration at scale.

#2. Incentives remain misaligned

That position is reinforced by incentives. Collaboration is encouraged, but it rarely sits within core supervisory expectations.

In other words, it exists alongside established obligations rather than within them. Institutions prioritize what is examined and enforced, and cross-border engagement does not consistently carry the same weight.

#3. Structures limit sustained collaboration

From there, the issue becomes structural. Existing models were not designed for continuous coordination across jurisdictions.

Public-private partnerships strengthen dialogue, but remain largely domestic. Meanwhile, private-to-private collaboration across borders is harder to sustain at scale.

So even where there is intent to collaborate, activity that spans institutions continues to be assessed in segments.

#4. Strategic and operational sharing remain separate

The separation between strategic and operational sharing adds another layer.

Strategic exchanges build understanding of patterns and trends. Meanwhile, operational exchanges support action on specific cases. These functions rarely connect.

As a result, context develops without timely action, or action takes place without broader context.

#5. The regulatory relationship itself remains unsettled

For institutions operating in the Mexico–U.S. corridor specifically, the regulatory landscape is active and evolving but not yet coherent.

In the United States, FinCEN has moved aggressively on corridor risk. In June 2025, FinCEN issued its first orders under the FEND Off Fentanyl Act, designating three Mexican financial institutions as primary money laundering concerns and imposing sweeping prohibitions on U.S. institutions’ ability to transact with them, a significant escalation of unilateral enforcement power. In Mexico, the CNBV responded quickly, assuming temporary management of the affected institutions to protect depositors and address the concerns raised by FinCEN.

That response illustrates something important. Bilateral engagement is possible when pressure is sufficient. But it remains reactive rather than structural.

On January 12, 2026, Mexico’s UIF co-hosted the first meeting of a new Transnational Organized Crime Working Group with FinCEN and other foreign partners, which is a notable departure from the UIF’s previous reluctance to engage in international cooperation on transnational security challenges. 

At the same time, the underlying tensions are visible. During the period of heightened enforcement pressure in 2025, Mexico’s UIF withheld AML data that it had previously shared through international channels, a reminder that political dynamics shape information flow in ways that institutional frameworks cannot fully anticipate.

The picture this creates is one of a corridor where regulatory engagement is intensifying, but where the infrastructure for sustained, structured collaboration between institutions across both jurisdictions does not yet exist. Enforcement action is not the same as intelligence sharing. And pressure applied at the regulatory level does not automatically translate into operational visibility for the institutions managing risk on the ground.

That gap between regulatory response and institutional intelligence is where private-sector collaboration becomes essential.

What does all this mean? The constraint lies in how collaboration is structured and supported.

So the outcome is consistent. Cross-border intelligence sharing remains uneven, and that unevenness defines the level of risk institutions carry across this corridor.

Two levels of intelligence banks must build simultaneously

Cross-border risk requires both context and timing.

Institutions need a clear view of how financial crime develops across jurisdictions, and they need to act while activity is still in motion. Most frameworks do not support both at the same time.

➡️Advanced intelligence: understanding the threat

Strategic intelligence builds a shared view of financial crime across jurisdictions.

It draws on typologies, patterns, and emerging risks, allowing institutions to track how activity evolves and where exposure is likely to build. That understanding shapes how risk is defined across the organization.

It informs model development, supports scenario planning, and aligns internal policies with external risk.

Without it, signals are assessed without sufficient context. Patterns are harder to recognize.

➡️Operational intelligence: acting on the threat

Operational intelligence focuses on action.

It provides signals that can be used while activity is still unfolding, allowing institutions to respond with greater speed and precision. Linked activity across accounts, counterparties, and jurisdictions can be identified as it develops.

Detection becomes more targeted. Investigations begin earlier. Network activity can be disrupted before it moves through the system.

It also sharpens alert prioritization, reducing noise and focusing attention on activity with broader relevance.

Where current approaches break down

These two forms of intelligence are often developed separately.

Strategic intelligence develops over time, but does not consistently feed into live detection. Operational intelligence supports immediate action, but often lacks broader context.

And that separation drives outcomes.

Institutions become well-informed and slow to act, or fast to act without a clear view of how activity connects across the system.

What is required

Both levels of intelligence need to operate together. Strategic insight needs to inform live detection. Operational signals need to be interpreted within a broader network context.

Because when those layers connect, institutions move closer to identifying activity as it develops across jurisdictions, rather than as isolated events within a single organization.

Public-private partnerships and the missing layer of cross-border collaboration

Public-private partnerships have strengthened how financial crime risk is understood and discussed.

They improve trust, support information flow, and create a shared view of emerging threats. In many jurisdictions, they are now an established part of the financial crime framework.

That progress is important to note.

Where PPPs are effective

• Build trust between institutions and authorities
• Improve information flow
• Strengthen shared awareness of threats and typologies

However, it isn’t the solution. Here’s why.

Where PPPs reach their limits

These models remain largely domestic.

Their effectiveness reduces when activity spans jurisdictions, particularly where coordination is required across multiple financial institutions.

PPPs are designed to connect institutions with authorities. They are less effective at enabling direct collaboration between institutions across borders.

The missing layer: private-to-private collaboration

Banks sit at critical points within cross-border financial flows.

Each institution sees only a portion of activity moving through the system. Viewed in isolation, that activity may appear routine. Viewed across institutions, it can form a coherent pattern.

And that view is difficult to establish without direct collaboration.

What private-to-private collaboration enables

• Pattern recognition across institutions
• Earlier identification of network activity
• Stronger, more contextual SAR development
• Early risk identification

Why this remains underdeveloped

• Legal uncertainty shapes what institutions are willing to share
• Data privacy considerations introduce additional constraints
• Limited infrastructure to support collaboration at scale

The case for Mexican institutions acting now

Regulatory hesitancy at the public level does not reduce the urgency for institutions operating in this corridor. If anything, it increases it.

The U.S.-Mexico trade relationship is one of the most significant economic relationships in the world. The corridor’s integrity underpins that relationship, and its disruption carries consequences that extend well beyond compliance risk. For Mexican financial institutions, maintaining trusted access to U.S. correspondent banking, payment infrastructure, and capital markets is not a peripheral concern. It is foundational to how they operate.

The 2025 designations of three Mexican financial institutions under the FEND Off Fentanyl Act made the stakes visible. Loss of U.S. correspondent banking access, regulatory intervention, and reputational damage are no longer theoretical outcomes. They are documented ones. And they signal that the standard of expectation for institutions operating in this corridor has moved, regardless of where formal regulatory coordination currently sits.

That creates a basis for collaboration that does not depend on regulatory instruction to exist. Mexican banks have a direct commercial interest in this corridor being well-managed. So do their U.S. counterparts. That shared interest is the foundation private-to-private collaboration can be built on.

Intelligence sharing opens the door data sharing cannot

Much of the hesitation around cross-border collaboration stems from conflating two distinct activities.

Data sharing, the transfer of customer records, transaction histories, and account-level information sits within a complex legal and privacy framework. The constraints are real, and institutional caution is understandable.

Intelligence sharing operates in different territory. Typologies, behavioural patterns, network characteristics, and emerging methodologies do not require the transfer of underlying customer data. Existing frameworks are considerably more accommodating of this kind of exchange, and the legal exposure is materially lower.

That distinction matters because it changes what is actually possible. Institutions do not need to resolve every legal question around data sharing to begin building shared understanding of how financial crime moves through this corridor. They can start with what they already have permission to share, and build from there.

The gap between what is legally possible and what is actually happening in this corridor is significant. Closing it does not require new frameworks. It requires institutions on both sides to recognise that the commercial case for acting is already strong enough to move.

What this means? Public-private partnerships remain foundational.

They do not carry the full burden of cross-border intelligence sharing.

Without effective private-to-private collaboration across jurisdictions, institutions continue to work with partial views of activity that are inherently connected.

Why now is different

These barriers are not new. The recognition that private-to-private collaboration is underdeveloped has existed for years, yet the gap has persisted. That requires an explanation.

The honest answer is that the conditions have not been in place to make it work. Legal uncertainty without safe harbor protection meant that sharing carried institutional risk with limited upside. Data privacy requirements made cross-border exchange difficult to structure responsibly. And without the technical infrastructure to enable collaboration without data transfer, the only viable model was one most institutions were unwilling to adopt.

What has shifted is not the intent. It is the feasibility.

Privacy-enhancing technologies and federated learning in particular remove the central obstacle. When institutions no longer need to transfer underlying data to generate shared intelligence, the legal and privacy constraints that previously blocked collaboration begin to ease. The exposure that made caution rational is no longer the inevitable cost of participating.

Regulatory posture has also moved. Supervisory expectations around responsible AI, model governance, and data protection have become clearer, which means institutions can now approach cross-border collaboration within a defined framework rather than in spite of one.

The barriers have not disappeared. But for the first time, the technical and regulatory conditions exist to work within them rather than around them.

Understanding what those conditions make possible starts with the technology itself. 

Technology as the enabler: a new model for intelligence sharing

Legal uncertainty, data sensitivity, and structural limits define how far collaboration can extend.

Working across institutions and jurisdictions has always meant balancing effectiveness with exposure, and that balance has shaped how far collaboration is taken, and where it stops.

Why banks need privacy-enhancing technologies

This is where privacy-enhancing technologies come in.

They support a different approach, allowing institutions to generate shared intelligence without transferring underlying data. Each institution retains control of its own information while contributing to a shared analytical outcome, and that opens up a different way of working across institutions.

Cross-border intelligence sharing has long been tied to data exchange. Where legal frameworks diverge and data protection requirements are strict, that link limits what institutions are willing to do. So when data no longer needs to move, that constraint begins to ease. This approach is essential in high-risk cross-border corridors.

Federated learning is one example.

Here, models are trained across multiple institutions while data remains local, which means insights can be shared while underlying data stays in place, and patterns can be identified across jurisdictions without exposing customer-level information.

Why this is relevant for the Mexico–U.S. corridor

Applied to the Mexico–U.S. corridor, the relevance is immediate.

Patterns that span institutions and jurisdictions can be identified without direct data sharing, and analysis can take place across institutions without introducing additional regulatory exposure.

At the same time, this aligns with supervisory focus on data protection and responsible AI, so institutions are not stepping outside existing expectations.

So institutions can develop and apply models across jurisdictions while maintaining control, transparency, and appropriate governance over how those models are used.

Instead of narrowing visibility to remain compliant, institutions can extend it while maintaining control of sensitive data.

What this enables

• ➡Cross-border pattern detection across institutions
• ➡Analysis across jurisdictions without transferring sensitive data
• ➡Alignment with data protection, model governance, and supervisory expectations (including responsible AI requirements)

What this changes

The link between collaboration and data movement begins to loosen, and that changes how institutions can approach cross-border intelligence.

Shared intelligence can be developed without transferring underlying data, within existing legal and supervisory boundaries.

Where this leads? From here, the direction becomes clearer.

Cross-border intelligence sharing begins to move beyond coordination and toward something more consistent, and over time it starts to operate with the characteristics of infrastructure.

Building a functional cross-border system

Cross-border collaboration has largely developed as good practice. In high-risk corridors, that position is no longer adequate, so it needs to operate as an expected capability.

Good practice is too optional. What this corridor requires is a different standard — one where cross-border intelligence sharing is treated as an expected capability, assessed as part of effective AML programs, and resourced accordingly.

That shift in expectation matters because behavior follows what is examined. When supervisors treat cross-border collaboration as peripheral, institutions treat it the same way. When it sits within core program expectations, participation becomes more consistent and the system begins to function differently. 

The difficulty is that regulatory signals and regulatory expectations are not yet the same thing. Enforcement action in this corridor has intensified. The message that financial crime risk here is a priority is clear. But that has not translated into explicit supervisory expectations around how institutions should be collaborating across it. The instruction to act has not followed the signal that action is needed. That gap exists across all three layers of coordination that this corridor requires — between institutions, between institutions and authorities, and between the authorities themselves. 

That gap will not close quickly. But institutions do not need to wait for it to close. The commercial case, the available mechanisms, and the technology to support them are already in place. The question is whether institutions on both sides of this corridor are willing to move ahead of the explicit expectation, because the cost of waiting is one that the system is already paying.

Cross-border financial crime already operates this way, so intelligence has to do the same.