Financial institution (FI) fraud and anti-money laundering (AML) teams are
finally starting to work together! 80% of them, in fact. Plus, 53% of institutions' AML and financial crime departments share part or all of the systems they use, and 100% of the FIs surveyed use artificial intelligence (AI) as part of their arsenal across
many different use cases.
What’s next? Reducing the total cost of ownership (TCO) of systems used by these related areas of the institution. Building not just a healthy return on investment (ROI) via such combinations, but also through gaining a ‘holistic’ view across both AML and
fraud data streams, and a ‘broader context for investigation’ to benefit the entire institution and its customers.
These are some of the most compelling insights provided by a recent survey of 30 US “mid-market” bank, credit union, and neobank anti-fraud and compliance experts, including a handful of IT and operations professionals as well. The survey, published in late
April by research firm Celent and
Hawk, a provider of AI-supported AML and fraud prevention technology, polled individuals from a mix of asset sizes mostly in the $1-20 billion range, though smaller and larger institutions were also included.
Don’t start with tech – start with people and processes
Discussing the survey with Hawk’s CEO and co-founder, Tobias Schweiger, who said that the first mistake many bank innovators and credit union change-makers make in their search for AI solutions for AML and fraud prevention and detection is to start with
technology. Instead, he suggested they “start with people and collaboration across processes.”
Anyone who works in the financial services industry or has read any of Finextra’s many articles on financial institutions’ ongoing efforts to prevent money laundering and detect and defeat fraud - or viewed webinars or attended conference sessions touching
on either discipline - know that this new collaboration trend is a revelation of sorts. That’s because many financial institutions - especially those of larger size – have over the past couple of decades built dedicated and proficient teams and systems to
manage KYC and client transaction screening. But they often intimated in interviews and conversations that they didn’t share many of their findings or data between departments.
Even with elaborate programs and systems in place to onboard the right customers and guard against the fraudsters - always on the attack to steal their customers’ money or damage their institutions’ reputations, or both – few institutions of any size, until
recently, have made it easy or even mandatory for their internal fraud and AML teams to communicate regularly and share helpful information with their counterparts. That’s changing, at long last, according to the responses shared from the 2025 Trends
in AML & Fraud Convergence at U.S. Mid-Market Banks survey.
Not always simple, but survey says collaboration between AML and anti-fraud teams works
Don’t call this new collaborative discipline 'FRAML', because 67% of respondents objected to the term, saying it “oversimplifies the complex processes” need to align fraud and AML team efforts in their institutions. Schweiger also pointed out that while
KYC and Extended Due Diligence (EDD) functions undertaken by banks are the most common areas of consolidation, case management and transaction monitoring are also now part of the mix.
“Combining processes allows information to flow more freely,” between back office functional areas, he explained. The ‘tech’ is better now, and many institutions now have comparable systems to handle both AML and fraud.
Many larger institutions have known and proven for decades just how helpful machine learning (ML) can be in examining large data sets to identify suspicious activity and unfamiliar transaction patterns. ML scripts and processes have been implemented to automatically
reduce the risk of human error, and perform repetitive tasks more efficiently than live staffers in many back-office departments. Schweiger says that by moving from reactive, after-the-fact detection to more proactive, predictive analysis, and leveraging the
unified data and greater context that a combined, or “FRAML” approach can provide, AI can more effectively pinpoint risks and reduce false positives.
Banks surveyed by Celent and Hawk ranked concerns around the cost of compliance, emerging types of financial crime, and increasing regulatory expectations highest in their responses to a question about the ‘top drivers of change’ in their fraud and AML programs.
However, close behind were keeping customer friction as low as possible and ongoing needs to support (evolving) digital financial services for their customers.
Top challenges for smaller FIs: Finding, keeping staff and reducing false positives
The top challenges – according to the survey - in doing all this? “Analyst staffing” and high rates of false positive alerts. The latter occur when people’s names, locations, or transactions trigger alarms or holds in processing because they’re deemed false
or fraudulent, and then those warnings turn out to be incorrect - consume valuable time for compliance staff and can lead to overburdened teams and the need to hire more to ‘cope with the workload.’
Schweiger said cross training of system users on both fraud and AML teams is becoming more important for all institutions. He also noted that AI can be a big help in solving the concerns and caseload burdens of staff while reducing false positives:
“AI is simply more precise in handling the complexity of cases,” he asserted, noting that modern behavioural biometrics software goes well beyond previous levels of scrutiny. “Generative AI has the ability to look back historically at transactions, segments,
and behaviours (of actors involved) and deviations can be seen clearly,” vs. past trends, Schweiger said.
By using AI to ‘define normal’ first, then proceed from that point into the future, bank systems can find anomalies, reduce false alerts, and delve deeper into specific data elements that might not be caught by a ‘hard and fast’ rule of screening, for example.
AI being used to supercharge data cleansing, speed investigations, and simplify report prep
After reducing false positives, AI comes into play most often in data cleansing, streamlining investigations, and automating ad-hoc web searches during such investigations, say its respondents.
It has also helped 27% of banks in writing their Suspicious Activity Reports (SARs) about certain questionable transactions. Many other useful and related fraud/AML applications are cited by more than 25-30% of institutions polled in the survey, so it’s
clear that AI/ML momentum is strong, and will likely continue to become stronger when forces are joined between protective departments and systems.
Financial impacts of AML/anti-fraud consolidation can be huge positive for institutions
There’s a big financial incentive to combining forces. Banks surveyed said they expected to save money – or had already realised impressive savings - from fraud and AML collaboration and consolidation of efforts as well. The figures cited ranged from hundreds
of thousands to millions of dollars per year in operating and technical costs. In fact, two-thirds of respondents that have already done so claim that consolidating their AML and fraud systems has saved them at least $1 million annually, and about 50% say
their expenses have decreased by more than $5 million per year.
The survey notes that sometimes, combination and collaboration efforts between fraud and KYC/AML teams don’t succeed, and while budget, staffing, training, and legacy technology can all be barriers to effective consolidation programs, the ‘cultural shifts’
and differences in focus and specific activities that prevent everything from working efficiently together – especially in the case of some previously ‘siloed’ departments - can also delay or depress the results of such initiatives.
Conversations are best start to find best anti-crime approach and technology for institution’s needs
Schweiger reiterated that both best practices and the survey results demonstrate the smartest way to ensure collaboration and consolidation between anti-fraud and AML functions’ achieve the goals desired by financial institutions is to begin with conversations
across departments, systems, and processes. Because, he explained, “if you don’t do that, you don’t know what technology change you want. People and processes first, then finding a targeted operating model (TOM) support it in the second step,” is the ideal
approach for banks and credit unions to take, he maintained.
While institutions have to match the technology they have and that’s become available recently to their specific needs, he pointed out that some banks and credit unions have found it valuable to begin with smaller pilot projects within AML and/or fraud prevention
departments. The “analytical work” performed in these initial studies can help identify cost savings and process improvements available from full implementation of proposed collaboration plans.
If KYC, AML, and anti-fraud staffers and systems don’t start working together more actively, employing new technology and innovations like AI, machine learning, and sharing data to make their combined functions more effective and more secure, there will
continue to be lots of wasted time and false alerts around transaction screening, onboarding, and other areas of scrutiny, said Schweiger.
Meanwhile, the criminals will keep hitting financial institutions and their customers from all sides, irrespective of which area or system is trying to prevent them from doing so, or even catch them in the act. It would be a real shame for the whole institution
as well as its customers, he said, as “you’ll miss financial crime by not working together – because fraud doesn’t consider AML and vice-versa.”