Since the Bank Secrecy Act (BSA) was enacted 54 years ago to prevent and detect money laundering and terrorist financing, the compliance obligation has evolved from a basic reporting requirement into a highly complex and continually evolving anti-money laundering (AML) regime.
An expert panel at A-Team Group’s RegTech Summit (NYC) in November convened to evaluate where industry readiness stands and how regulatory technology (RegTech) and generative AI (GenAI) are helping to address these dynamic challenges.The panel included Alex Ford, Co-Founder, Women in RegTech New York, Moderator; Krishna Nair, Head of AML, KYC, Electronic Communication, Trade Compliance Technologies, MUFG; Joseph Sabelja, Generative AI Firmwide Platform Lead, JP Morgan and Chandrakant Maheshwari, FVP, Lead Model Validator, Flagstar Bank.
The session kicked off with a live audience poll asking, “Which of the following do you believe will have the greatest impact on addressing financial crime and AML compliance in the next two to three years?”
As the audience was voting, the discussion focussed on the current state of AML compliance, with one panelist observing, “FinCEN used to go after the bank, now they’re going after the chief BSA officer,” and noting “Two areas that didn’t make enough progress are dealing with the foundational issue of data, and the technology itself is not agile enough.”
A review of the audience polling indicated clear plurality (45%) selecting “Harnessing data to build a single, unified view of the customer,” with “Leveraging AI and generative AI” at 30%. The remaining votes went to “Keeping up with regulatory changes” (15%), “Achieving perpetual KYC” (10%), and “Automating KYC/AML processes with RegTech solutions” (0%).With the poll validating data as a primary concern, another expert cautioned against letting data imperfections delay innovation, suggesting that “we always have to be careful about putting too many impediments in the GenAI journey” suggesting instead that firms “expose those issues to the users,” and making incremental progress with appropriate use cases.
Beneficial Ownership
The discussion then switched to the Corporate Transparency Act of 2021 (CTA) and the Financial Crimes Enforcement Network (FinCEN) Beneficial Ownership Information (BOI) reporting rule changes and how financial institutions plan to integrate data from the proposed new registry into their compliance processes.
One panelist described the registry as “a wonderful concept,” while cautioning that challenges related to privacy, jurisdictional differences, and verifying the accuracy of data remain. Despite these hurdles, this speaker expressed optimism, observing that the registry is “a step in the right direction” and should ultimately “reduce the friction” between clients and financial institutions.
Another panelist raised uncertainty over whether using the registry would actually indemnify the firm and simplify compliance, or simply add to existing due diligence burden. “If the FinCEN says this is the authoritative source, we are in good shape,” but warned that if it is “just guidance,” the resulting overlap could be more costly.
Editorial note: The CTA has faced several legal challenges in the US courts and at the time of writing, enforcement of the CTA remains suspended until a definitive judicial resolution is reached, leaving the future of BOI reporting and any potential benefits in a state of flux.
The potential of GenAI
The panel next turned to the role of generative AI in tackling financial crime. One expert described how it builds on decades of advances, remarking that “this is evolutionary, not revolutionary.” GenAI can augment existing rule-based and predictive AI systems rather than replace them and expands the ability to recognize patterns across text, voice, or images, offering new levels of automation in areas like suspicious activity reports (SARs).
GenAI can gather and summarize vast amounts of open-source data, connect external sources such as multiple news vendors, and then integrate the findings into a graph database. In this view, combining data from multiple channels into a single structure not only enhances detection but also helps produce more accurate reports. This expert stressed that fully harnessing these capabilities may still be “a long way … at least five years” in coming, yet the potential to streamline large-scale data analysis remains significant.
Perpetual KYC and Single Customer View
Panellists agreed that achieving a single, holistic view of each customer is central to improving financial crime detection. But one panelist stressed the challenge of balance the costs of running advanced technology in parallel with existing systems until “we can prove … it is as good, if not better.” Firms will need to demonstrate the reliability of any new AI or data solution to gain “regulatory trust.”.
Turning to the concept of perpetual KYC (know your customer), a panelist described it as “a fancy term invented by the consulting companies.” Rather than wholly reinventing KYC processes, several panel members advocated making them seamless within normal business workflows. As one put it, “KYC is an enterprise issue,” suggesting that effective solutions require collaboration across departments—credit, risk, and compliance—and more strategic use of data already in hand.
The panel emphasized that cultural change is needed to reduce friction for clients. One noted that by shifting the mindset around KYC reviews to a broader business engagement, firms can improve service while still meeting their compliance obligations. Panellists were keen to highlight that meaningful innovation in KYC can serve both revenue and regulatory objectives, especially if supported by tools for real-time monitoring and by a shared understanding of each customer’s complete profile.
Finally, an expert gave an example of how the integration of GenAI with graph databases can offer benefits beyond compliance. “Okay, so a new story comes in, Elon Musk is now the head of DOGE (Department Of Government Efficiency) … all of a sudden, you trigger an alert that says, he’s a PEP (politically exposed person). That goes up to one department. Or maybe Joe does a leverage buyout of 7-Eleven and all of a sudden, this is a sales opportunity for the firm. Once you have the data…then you can expose it in many different ways.”
The panel session concluded with an audience question asking, “How do you think SARS may change with AI adoption and evolving cybercrime schemes?”. One panelist noted that manually SARs can be “terrible, full of spelling mistakes … I pity the poor humans who have to read this stuff…I look forward to the day when we get rid of the SARs.” indicating that over time, automated mechanisms can reduce repetitive data entry and lighten workloads.
Another speaker reinforced this point by pointing to opportunities for “automation and generation,” suggesting that seamless data flows could eventually remove the need for manual retyping of details across multiple systems. These comments reflected a broader panel consensus that GenAI, particularly when integrated with existing compliance processes, can produce more complete and accurate filings. In doing so, it could free human investigators to focus on higher-value tasks, without sacrificing the oversight essential for regulatory adherence.
Subscribe to our newsletter
