Most banks were built to combat fraud and money laundering as two separate functions, with separate teams and goals. The fraud team aims to prevent the bank and its customers from losing money, often in real time during a payment transaction. On the AML side, the goal is to detect and report suspicious activity to satisfy the regulators, often well after the fact. Different tools, data, and reporting lines mean that, as ACAMS has pointed out, in many institutions, the fraud group and the AML group don’t communicate or collaborate. FRAML, short for Fraud and Anti-Money Laundering, is the attempt to bridge that gap and address financial crime as one connected issue instead of two.
This approach is based on simple logic when considering how criminals actually operate. Fraud is a predicate offense for money laundering. Fraud is one of the underlying crimes that generates dirty money. It creates the money, then money laundering moves, layers, and hides it. So where there is fraud, money laundering is mostly not far behind.
Fraud now features in 89% of FATF mutual evaluations, making it the second most common predicate offense after corruption, according to the FATF Annual Report 2024-2025.
Criminals do not respect organizational charts. A single criminal network can conduct APP fraud, move proceeds through mule accounts, and structure transactions to avoid reporting thresholds, all in one coordinated operation. If each piece is kept in its own silo, a bank only ever sees part of the picture.
A few forces have taken this from a good idea to something companies can’t ignore. Instant payments have reduced the detection window to seconds, and the European Banking Authority has warned that the risk of fraud in instant payment settings is ten times higher than in traditional transfer channels. AI is driving deepfakes, synthetic identities, and scams at a pace and scale never seen before. The sums involved are vast: UK criminals stole more than £600 million in the first half of 2025, and Deloitte estimates US authorized push payment fraud losses could reach almost $15 billion by 2028. FinCEN has been encouraging banks to improve communication and collaboration between their AML, fraud prevention, and cybersecurity units, and supervisors are increasingly judging a program on whether it actually works.
FinCEN has encouraged this kind of internal information sharing since 2016
FRAML is less about creating one merged department for your compliance program and more about connecting what each side already knows: Shared data, a common view of customer risk, and signals that get to the other team at the moment they matter.
The following topics are going to be covered in this article;
- What Is FRAML?
- Why Fraud and AML Are Converging Now?
- FRAML vs the Siloed Model
- The Benefits of Convergence
- The Challenges
- What Convergence Looks Like in Practice?
- How to Start Converging?
- FAQ
1. What Is FRAML?
Fraud and Anti-Money Laundering, or FRAML, is the integrated approach to the fraud detection and AML functions within a financial services organization. FRAML doesn’t run separate teams, separate data, and separate tools. Instead, it shares signals across both to catch financial crime that siloed systems often miss.
2. Why Fraud and AML Are Converging Now
The pressure to converge is mostly practical, and it has been building for a few years. Start with cost and staffing. Running two separate stacks means duplicate systems, duplicate alerts, and two teams investigating versions of the same case. Mid-market banks feel this hardest, since they have the smallest budgets and the thinnest benches, and financial crime talent is scarce and expensive. The savings are real, it is reported that 77% of industry professionals expect to save over $1 million within five years of converging, and two-thirds point to improved operational efficiency as a benefit.
There has been an undeniable shift to real-time rails. Instant payments don't wait for a Monday-morning case handoff. When money leaves an account in seconds and can't be recalled, a fraud analyst who spots something has no time to email the AML team and hope someone picks it up. The detection has to happen in one connected motion or it doesn't happen at all.
The money mule is where the two disciplines physically meet. A scam victim's payment lands in a mule account, and at that exact moment the problem stops being "fraud" and becomes "laundering." Same money, same account, same network, just viewed by two different teams. The US financial institutions detected a big spike in money laundering accounts in early 2025. Many of these scam accounts proceed by moving through mules. Treating that seam as two separate problems is how criminals slip through.
Regulators are leaning the same way, asking for stronger results and tighter cross-team coordination instead of two programs that never compare notes. The direction is clear. In Celent's 2025 survey of US mid-market banks, 53% planned to increase consolidation of AML and fraud, with 40% currently in the process of converging systems, which works out to roughly 93% actively pursuing or planning convergence. No institution surveyed showed no interest in making the shift.
3. FRAML vs the Siloed Model
The siloed model splits one criminal chain across two teams that don't share much. FRAML keeps that chain in a single view. Here's the side-by-side comparison table;
|
Dimension |
Siloed Model |
FRAML (Converged) |
|
Team structure |
Fraud and AML sit in separate departments with different reporting lines and little day-to-day contact |
Shared leadership or one financial crime function; teams coordinate closely or sit together |
|
Data |
Separate systems and data sets, so each side sees only its own slice of the customer |
Shared data feeding a single, 360-degree view of customer risk |
|
Detection logic |
Fraud works in real time (authorization rules, device signals, velocity limits); AML reviews patterns after the fact (transaction monitoring, customer due diligence) |
Fraud and AML signals feed each other, pairing real-time blocking with pattern analysis |
|
Case management |
Separate queues and tools; cases handed off manually, if at all |
One case management system both teams use, so context travels with the case |
|
Cost |
Duplicate platforms, duplicate alerts, duplicate investigations of the same activity |
Less duplication and lower total cost of ownership; reported savings often $1M+, sometimes $5M+ |
|
Detection gaps |
A fraud case closed internally may never reach AML or trigger a SAR; mule and laundering links go unseen |
Connected signals surface the full chain, from scam to mule account to laundering |
Table 1: FRAML vs Siloed Model
A fraud case resolved internally, without escalation to BSA/AML, may never trigger the SAR review regulators require. That is the blind spot convergence is meant to close, by giving both teams a 360-degree view of customer behavior and suspicious activity instead of two half-pictures.
4. The Benefits of Convergence
The main benefits of convergence are that it eliminates blind spots, it reduces operating costs, and it accelerates response.
Fewer blind spots:
This is the benefit that is hardest to put a dollar figure on and the one that matters the most. When fraud and AML share data, signals that one team might have shrugged off suddenly mean something to the other. A fraud alert that is closed as a one-off appears very different when considered alongside a pattern of mule activity that the AML side is already tracking. That is exactly where the criminals have been hiding. The entire chain connects two surfaces.
Lower total cost of ownership:
A single converged stack is more cost-effective than running two. The report also found that 77 percent of respondents expect to save more than $1 million in the first five years of convergence, with 36 percent expecting savings to exceed $5 million. More than half mentioned lower total cost of ownership as a benefit. Savings are made by removing duplicate systems, duplicate licenses, and duplicate investigations of the same activity.
Faster response time:
Improved operational efficiency was the main benefit of convergence, according to two-thirds of institutions surveyed. Shared case management means the context travels with the case, rather than getting lost in a handoff, so investigators spend less time rebuilding what the other team already knew. AI helps here, too: 57% of banks were using AI to reduce false positives, freeing up analysts to work on the alerts that matter, rather than chasing noise.
The throughline is the benefits they present to each other. Better data means fewer false positives, fewer false positives mean faster cases, and faster cases mean threats get caught while the money is still in reach.
5. The Challenges
Convergence sounds promising on a slide. It’s a messy process, and even those who study it concede the label overstates the work. As Celent notes, the term "FRAML" is often used as a shorthand for a complex and multi-layered transformation. The following five obstacles are encountered repeatedly:
Aging systems:
Most banks built fraud and AML as separate systems, so the platforms can’t talk to each other without heavy plumbing. The timing is especially tricky: Fraud detection is real-time, but much AML monitoring is done in overnight batches. Merging real-time and batch processes into one flow is real engineering, not a config change.
Data silos:
Fraud data and AML data are frequently kept in separate silos, with different formats, different owners, and different definitions for the same field. You can’t get to a “single view of the customer” until somebody has cleaned and mapped and merged all of that. Celent says the main things holding back the path forward are operational and data silos, aging systems, and the initial investment.
Different key performance indicators:
The two teams have different metrics. Fraud and AML have different business aims, processes, procedures, and KPIs. Fraud is measured by losses prevented and speed. AML is measured by regulatory compliance and quality of filings. Those goals do not always align. A converged program has to make peace with those goals, not just let one win quietly.
Upfront investment:
The first thing is the spending and the second is the savings. Making the business case to start with is one of the more difficult steps, and new technology, integration work, and retraining all cost money before any return lands.
Change management:
Change management is often perceived as the less challenging aspect, but it can lead to the failure of difficult projects. You’re asking two departments with different cultures, skills, and incentives to share tools, queues, and occasionally a manager. People are protective of their turf, and new workflows are resisted. Without leadership buy-in and a clear plan, even perfect tech can end in a stalled rollout.
None of these factors is an excuse for avoiding convergence. They're why you run it as a staged program with executive sponsorship, not a switch you flip.
6. What Convergence Looks Like in Practice
After removing the strategic discussions, convergence refers to four specific mechanical changes in the actual workflow:
Shared signals:
Rather than two teams drawing from two narrow feeds, the two teams draw from one pool of inputs. Modern setups blend customer, transaction, device, behavioral, and third-party intelligence into one layer. A device fingerprint that trips a fraud rule is now visible to the AML side, and a sudden change in transaction behavior that interests AML is visible to fraud. The same evidence does both jobs, rather than sitting locked on one side of the house.
A unified risk score:
Rather than having two separate scores for the same customer, the platform merges fraud and AML into one dynamic risk perspective. This process typically refers to master data management, creating a single record for a customer, account, and transactions, so the score is based on everything known about a person, not just what one module happened to see.
One case queue:
Both domains’ alerts are fed into a single case management system. The system provides the team with a 360-degree view of customer behavior and suspicious activity. An investigator opens a case and sees the full customer in one place. The case moves with the context, so nothing is lost in a handoff between teams.
Cross-module analysis:
This is what siloed tools can’t do. The system compares a fraud signal to laundering patterns and vice versa, discovering connections that neither module could uncover independently. Tools like dynamic dashboarding, scenario testing, and behavioral intelligence offer decision-makers a single risk view in line with standards such as FATF, GDPR, and the EU AML package.
The flow is as follows: Signals are consolidated into a unified score, which then directs alerts into a single queue, and cross-module analysis connects the dots into a cohesive narrative. That’s the practical difference between two programs that sometimes email each other and one that really does operate as a whole.
The unified risk score model serves as the foundation for the Sanction Scanner's AI-powered FUSION platform. Fusion is a unified platform for fraud and compliance that combines all module outputs into a single risk perspective, computes a dynamic customer risk score, and applies it to all subsequent checks.
7. How to Start Converging AML and Fraud
The mistake most teams make is to try to merge everything all at once. One better way is to do it as a sequence. Each step allows for the next step. First, it’s helpful to do a maturity assessment of where you are today in terms of integration, data sharing, typology alignment, case management, governance, and technology architecture. Instead of guessing, it tells you where the real gaps are.
Step 1: Create a shared data layer: This is what everything downstream depends on. As long as fraud data and AML data sit in separate stores with different formats, no amount of process change will give you one view of the customer. The job is, first, to bring those sources into unified records across customers, accounts, and transactions so both teams are working off the same facts. This is the slow, unglamorous part, and it’s also the part that makes or breaks the rest of it.
Step 2: Typology matching: After sharing the data, ensure that the two teams communicate using a common risk language. Find out where the fraud and laundering patterns intersect and agree on the meaning of a given signal to each side. A practical way to get started is to identify the top three typologies in your portfolio that would benefit the most from fraud and AML signal correlation to improve detection and to pilot a unified detection approach for those typologies. Start small so you can prove value and learn before you scale across the book.
Step 3: Unify case management: Feed alerts from both teams into one case system with shared data and aligned typologies. Now, when an investigator opens a case, they can see the complete customer information, and the context remains with the case instead of being lost during handoffs. This process is also the step that yields visible efficiency gains, as people stop rebuilding what the other team already knew.
If the pilot is successful, factor the results into your broader plan and revisit your financial crime technology roadmap to ensure convergence is baked in. Maintain executive sponsorship throughout. Much more often convergence stalls on people and priorities than on technology.
Sources
[1] Celent. Trends in Fraud and AML Convergence at US Mid-Market Banks & Credit Unions. April 2025.
[2] ACAMS. Financial Crimes Convergence: The Case for Integrating Fraud and Anti-Money Laundering Processes.
[3] Financial Action Task Force. FATF Annual Report 2024-2025. 2025.
[4] European Banking Authority. Opinion on New Types of Payment Fraud and Possible Mitigants. 29 April 2024.
[5] UK Finance. Half Year Fraud Report 2025. October 2025.
[6] Deloitte Center for Financial Services. The Rise of Authorized Push Payment Fraud. October 2025.
[7] Financial Crimes Enforcement Network. Advisory to Financial Institutions on Cyber-Events and Cyber-Enabled Crime (FIN-2016-A005). 25 October 2016.
FAQ's Blog Post
It presents real governance issues, which is why data governance has become central to convergence. You have to follow rules like FATF, GDPR, and the EU AML package to merge customer, transaction, and behavioral data into a single view. The answer isn’t to keep the data separate but to build the governance layer in parallel with the shared platform so that access and use are still controlled.
No, FRAML is not only for big banks. The reality is the opposite, in fact. Large banks often face the greatest challenges due to deeply siloed business lines and complex system architectures. Celent's take is that mid-market banks are best positioned to benefit from the convergence of AML and fraud detection, since they run leaner and feel the pressure of duplicated costs and staffing most acutely. Smaller institutions are often more capable of converging.
They're different stages of the same problem. Fraud is the crime upstream creating proceeds; AML targets the concealment of those proceeds. Fraud owns real-time controls (authorization rules, device signals, etc.), and AML owns monitoring, due diligence, and SAR filing. The problem is not all AML cases are initiated through fraud. Structuring, sanctions violations, and terrorist financing do not directly hurt the customer, so the two overlap instead of being a straight line.
Begin with data, not reorganisation. Build a shared data layer, as nothing else works without it. Then align the typologies where fraud and laundering overlap, pilot your top few high-risk patterns, and only then unify case management. A maturity assessment up front will reveal your real gaps. Keep executive sponsorship attached throughout, because convergence stalls on people much more than on technology.
Either or both, depending on the institution. There is not one model of convergence. It is a spectrum. Some banks fully merge the departments; others maintain two teams but link them with shared data and a common case system. It's not about a particular org chart. It’s getting the intelligence each side produces to the other when it matters. Many begin with shared systems and let the team structure emerge.
No, FRAML is not required. There is no rule that says you have to merge fraud and AML. But regulators obviously like the direction. FinCEN has urged banks to increase communication and cooperation across their AML, fraud prevention, and cybersecurity teams, and supervisors are increasingly evaluating programs on whether they actually catch crime. So not a legal requirement, but an expectation and best practice.
