AI vs Financial Crime: Securing Real-Time Cross-Border Payments

Cross-border payments are becoming faster than ever. So is financial crime.

The same digital evolution that has enabled faster legitimate payments has also enabled faster illicit ones. Fraud rings, mule account networks, and synthetic identities are moving through payment ecosystems with a speed and sophistication that manual compliance processes were not built to match.

This is creating a new race across the payments industry as it works to stay one step ahead of bad actors. As the window for identifying, investigating and stopping suspicious activity continues to shrink, payment providers, financial institutions and networks face growing pressure to detect risk without slowing legitimate transactions.

The response cannot be more compliance and slower processing with greater rigour. Compliance in payments has to operate at the same velocity as the payment itself, and that requires a different approach to how risk is detected and prevented across cross-border flows.

This piece looks at how financial crime is evolving alongside payment infrastructure, why traditional compliance frameworks are under pressure, and how AI and network-level controls are changing what effective prevention looks like.

Table of contents 

Cross-border payments are moving faster than ever

The time it takes for money to move between markets has been reduced from days to minutes, and in many cases, flows are now near-instantaneous. 

Several trends are driving this shift:

• Real-time payment infrastructure is live across a growing number of major and emerging markets, enabling instant account-to-account transfers that bypass traditional clearing windows
• Mobile wallets have become a primary channel for cross-border payments, extending reach into markets where bank account penetration is lower
• Stablecoin-based liquidity models are supporting payment activity outside traditional banking hours, removing timing constraints that previously slowed cross-border flows
• Cross-border payment services are increasingly available around the clock, meaning settlement no longer depends on overlapping business hours between markets

For businesses and consumers, these developments create a better payment experience. Money can move more quickly, reach more destinations across more markets and currencies, and support a broader range of use cases.

The compliance implication runs in the opposite direction. 

Every reduction in settlement time is also a reduction in the window available to review, investigate, and stop suspicious activity. A transaction that previously moved through multiple institutions over several days may now be completed in minutes, leaving far less time for intervention when something looks wrong.

As payment volumes increase and payment ecosystems become more interconnected, the demands placed on compliance teams grow even faster. The time available to act is shrinking, while the amount of activity that must be monitored continues to expand.

Financial crime is becoming faster, too

Illicit financial flows have surged by almost 20% annually, reaching an estimated $4.4 trillion in 2025, a rate that reflects how effectively criminal organisations are exploiting digital payment ecosystems.

The methods driving that growth share a common logic: distribute activity across enough accounts, institutions, and jurisdictions that no single participant can see enough to act.

Several capabilities are making that easier:

• Fraud rings operate across multiple jurisdictions simultaneously, distributing activity in ways that make it harder to detect through single-institution monitoring
• Mule account networks move funds through multiple accounts in rapid succession, creating layers between the origin of funds and their destination
• Synthetic identities and digitally fabricated credentials make it harder to verify that the person initiating a payment is who they claim to be
• AI is giving criminal networks capabilities and automation to increase the scale and speed of fraudulent activity that were previously out of reach

That last point deserves particular attention. AI is not just making existing financial crime methods more efficient; it is enabling new ones.

Deepfaked credentials, AI-generated correspondence, and adaptive fraud patterns that learn from detection attempts are creating a threat landscape that looks materially different from the one compliance frameworks were originally designed for.

The compounding challenge is that these activities seldom sit within a single institution or payment flow. A financial crime event may span multiple accounts, payment methods, jurisdictions, and counterparties, with each participant seeing only a fragment of the broader picture. 

Risk that appears insignificant in isolation can be part of a pattern that becomes visible only when transaction activity is viewed across the network as a whole.

Why compliance needs to operate at network speed

The challenge facing compliance teams is complicated because the volume, speed, and distribution of payment activity have outgrown the operational model that most compliance frameworks were built around.

The limitations of traditional compliance approaches

Traditional compliance frameworks were designed for environments where transactions moved slowly enough that investigators had time to review flags, request documentation, and escalate cases before funds settled. That window has largely disappeared in real-time payment environments.

Rules-based monitoring still plays an important role. But static rules applied to individual transactions struggle to detect activity that is deliberately distributed across multiple accounts, institutions, and jurisdictions to avoid triggering any single alert. 

Why AI changes the equation

The same properties that make AI useful to criminal networks – the ability to process large amounts of data quickly, identify patterns, and adapt to new information – make it a meaningful tool for compliance teams tasked with countering them.

Where rules-based systems flag individual transactions against predefined criteria, AI can:

• Analyse behavioural patterns across payment flows and disparate sources of information
• Identify connections between accounts and counterparties that would not be visible through manual review
• Surface higher-risk activity for investigation without requiring analysts to process every alert manually

In cross-border payments specifically, where a single financial crime event may span multiple markets, payment methods, and institutions, the ability to build a connected picture across fragmented data is where AI adds the most value. 

It doesn’t replace compliance teams, but it does give them the analytical reach that the scale and speed of modern payment flows require.

Collaboration is key

Addressing the challenge of financial crime also requires cooperation across the payments ecosystem. 

No single institution sees the entire payment journey. Stronger information sharing, coordinated compliance frameworks, and network-level controls give participants a clearer view of risk than any one of them could achieve independently. 

This creates an important role for global payment networks like Thunes, which help connect participants across markets while embedding the compliance controls they need to facilitate secure cross-border transactions.

Building compliance for the next generation of cross-border payments

Modern payment networks must balance two priorities simultaneously: supporting the efficient movement of money and maintaining the controls needed to identify and manage financial crime risk.

Thunes approaches this as a network-level challenge rather than a per-transaction one. 

The Thunes Direct Global Network 

Through the Direct Global Network, compliance controls are embedded across payment flows, giving participants consistent oversight across markets without the need to build and maintain separate compliance frameworks. 

A single connection provides access to multiple markets under consistent standards, reducing the fragmentation that makes cross-border financial crime difficult to detect and prevent.

Fortress Compliance Platform

Fortress Compliance provides the foundational layer needed for security in cross-border payments. It underpins all transactions in the Direct Global Network. 

• KYC and due diligence frameworks support participant verification at onboarding and via ongoing risk management processes. 
• Transaction monitoring operates continuously across payment flows, surfacing potentially suspicious activity for review rather than retrospectively. 
• Sanctions screening controls apply across multiple markets and are aligned with the regulatory requirements of each jurisdiction the payment touches.

This provides real-time compliance to real-time payments. 

AI-powered detection through Tookitaki

Thunes’ partnership with Tookitaki extends these capabilities into AI-powered financial crime detection.  

1. The Anti-Money Laundering Suite applies machine learning across transaction data to identify behavioural patterns, connections between accounts and counterparties, and activity that falls outside expected parameters at a scale and speed that rules-based systems cannot match.
2. The Reconciliation Suite applies the same AI-enabled approach to reconciliation, reducing the manual overhead typically generated by high-volume cross-border payment flows.

Together, these capabilities address the core challenge posed by the pace of modern payments: the need to detect and respond to financial crime risk at the same speed at which payments move, across a network where risk signals are distributed among multiple participants, markets, and payment flows.

Cross-border payments will continue to get faster. Criminal networks will continue to adapt. The infrastructure that sits between them needs to do both, maintaining the controls that keep money moving securely without becoming the bottleneck that slows it down. That is the balance Thunes is built to support.

To see how AI-powered AML could help you, speak to one of the team.