A deep dive into how AI is redefining compliance strategies and reshaping financial crime risk management at banks.
To start off, can you please share a bit about your professional journey—your experience at major banks and what led you to your role at WorkFusion?
2025 marks the beginning of my 30th year in financial crime compliance, with much of my experience focused on building and managing Anti-Money Laundering (AML) and sanctions compliance programs. I have served as the Chief Compliance and AML Officer at a national bank (Riggs) and later became the CEO of a consulting firm from 2005 to 2015, where I assisted dozens of U.S. and international banks in responding to regulatory enforcement actions and Department of Justice investigations involving AML issues. In this role, I led a team that remediated AML compliance failures and then built sustainable AML operations. In 2015, I transitioned from consulting to the AML and financial crime compliance software industry, where I directed all client and customer initiatives for a Regtech startup that helped banks improve how they find and use Risk Intelligence Data, such as sanctions, watchlists, Politically Exposed Person records, and Adverse Media. My role at WorkFusion is to assist each of our core functions: Product in creating better products for our AML users; Sales in having more effective conversations with AML buyers; and Marketing in spreading the message of what we can do to strengthen AML compliance.
AI is playing an increasingly critical role in banking compliance. From your perspective, how has AI transformed areas like anti-money laundering and risk management?
AI is transforming AML and risk management in numerous ways. Among the most impactful changes is the automation of much of the routine, time-consuming, and error-prone data gathering and information collection work. The primary objective of AML and risk management is to make sound decisions that protect financial institutions while promoting strong business practices. The challenge is that for decades, AML and risk management teams have spent more time gathering and assembling the necessary information to make decisions than actually making them. This situation is frustrating for workers and risky for institutions. AI and its associated technologies not only expedite information gathering but also incorporate additional data sources, doing so with a level of consistency that often escapes individuals whose focus wanes as they click, copy, and paste all day.
False positives have long been a major pain point in AML programs. Why do you believe AI is finally making it possible to solve this issue effectively?
Our take on false positives is this: AI makes them irrelevant. To many in the AML field, this might sound radical. For the past decade, a significant amount of time, effort, and money have been spent on “reducing false positives”; it’s ingrained in an AML executive’s mindset. However, this is no longer necessary. Why? False positives primarily drain human and financial resources from AML teams. If alerts are not reviewed and resolved promptly, backlogs grow, and regulatory filing deadlines are at risk. The traditional solution has been to hire more personnel or outsource the work. AI changes this because, with AI, every alert — even those we have called “false positives” — can be reviewed in seconds. When tasks that once took minutes or hours can now be reduced by 95%, just let AI handle the alerts.
Can you explain how financial institutions are using AI to streamline and modernize their compliance operations?
In addition to automating data and information gathering, AI is becoming a worker’s assistant in this sense: it organizes and analyzes the risk data needed for workers to make informed decisions. We believe that humans should remain the ultimate decision-makers, but AI can now present information in ways that make decisions more informed. This results from freeing up time for individuals to slow down and focus on important issues. Additionally, AI can be programmed to assist in decision-making, offering suggestions or recommendations along with supporting evidence and written rationale that the worker can review and validate. For example, in AML, analysts are buried in sanctions and adverse media alerts. They have aggressive “production” requirements to review these alerts fast. Unsurprisingly, this can lead to errors, which increases risk. AI can now review these alerts using the same procedures that workers currently use, present a written summary, and offer a decision to either close the alert or escalate it for further investigation. The analyst can now quickly review the decision recommendation, scan the AI-generated written explanation, and move on in a fraction of the time it took before AI.
“AI agents” is a term we’re hearing more and more in banking—what exactly are AI agents, and how are they impacting day-to-day compliance functions?
An “AI agent” is a digital co-worker that makes decisions, acts, and communicates. Think of an AI agent as a trained and experienced new member of an AML team who joins an organization with five years of experience in their specialty. For example, they could be a specialist in KYC and onboarding risk assessment, a sanctions screening alert analyst, a Transaction Monitoring analyst, or an Enhanced Due Diligence case investigator. They come with the experience of having reviewed, resolved, and decided thousands of matters and need to spend a few days, or perhaps a week, learning the nuanced procedures of a new institution. AI agents are transforming the way compliance teams operate. For one, many of the roles they fill tend to have high turnover due to the repetitive nature of this work. Financial crimes management recognizes that many of the jobs can be monotonous. People often become bored, which leads to dissatisfaction. The result of that is increased replacement costs and, perhaps worse, increased risk as work backs up.
In what ways are AI-powered tools improving accuracy and efficiency in identifying suspicious activity, compared to traditional rule-based systems?
We are still in the early stages of using AI to replace decades-old rule-based transaction monitoring systems. This approach makes sense because, while we strongly support AI adoption, transaction monitoring systems are the backbone of AML compliance. Changing how suspicious activity is identified will take time, as institutions need to ensure AI methods match or exceed the capabilities of current rule-based systems. There will also be increased regulatory scrutiny in this area, as the risks are significant for both institutions and regulatory agencies.
That said, it is important for AML leaders to understand how Machine Learning (ML) and Deep Learning will transform suspicious activity detection. Briefly, Machine Learning involves algorithms that analyze data to recognize patterns and make predictions or decisions, often requiring human-designed features. This includes techniques like supervised learning and reinforcement learning. Machine learning models rely on structured data and typically require manual feature selection (e.g., transaction frequency, amount) to perform effectively. It might be more accurate to say Machine Learning involves more Machine Teaching.
Deep Learning is a subset of ML that uses neural networks with multiple layers (hence “deep”) to automatically learn features from raw data, guided by human input. It excels at processing unstructured or complex data and large data volumes. Deep Learning is made possible by increased computing power driven by AI chips. Although wide adoption by AML teams may still be a few years away, once implemented, it is likely that the amount of suspicious activity identified will grow exponentially, with significant secondary impacts on the AML workforce.
How do you see AI balancing innovation with regulatory expectations, particularly in a highly scrutinized space like financial crime?
Keep in mind that every regulatory agency has issued statements encouraging financial institutions to explore and, when appropriate, adopt new technology, including AI. However, this encouragement comes with a warning: ensure that decisions about which AI to implement and how to deploy it are carefully considered, properly managed, and include sufficient human oversight. This is often referred to as “explainability.” Many in the financial crime compliance field now question what exactly “explainability” means. We believe the definition is simple — financial crime management must be confident in understanding AI and be able to explain how it works, along with the steps their institution took to decide on and implement it. If AML leaders are not fluent in the language and operation of AI, they need to become proficient as soon as possible. We aren’t suggesting AML practitioners need to be technical or engineering experts. However, if they lack the ability to explain it simply, then that means they have more to learn.
What’s your personal strategy when it comes to helping banks adopt and implement AI-driven compliance solutions successfully?
At this point, if banks are not already deploying AI compliance applications, it suggests they see the risks of implementation as greater than those of non-implementation. However, we believe the risk of non-implementation is significantly greater at this stage. We understand that, generally, the idea of AI can be intimidating, causing hesitation. In the past, taking a wait-and-see stance made sense to watch how others fared and what hurdles they encountered, including regulatory concerns. When software wasn’t as transformative, this strategy was probably wise, giving time to catch up. But, with AI’s rapid pace of improvement in AI, peer institutions can quickly improve and make significant progress, building momentum fast. Falling behind now has more serious consequences. No institution wants to be an outlier among its peers, as that raises concerns with regulators, who may then take aggressive measures. So, when we see banks waiting and observing, we try to understand their fears, address them, and help them take small, manageable steps to implement low-risk AI solutions. This allows them to see firsthand how AI works and the benefits it can bring to operations, costs, and compliance.
For compliance and risk professionals looking to explore AI, what’s one piece of practical advice you’d offer as they begin this journey?
Learn. Specifically, start reading about AI, and understand that the term “AI” is broad. Explore AI’s component technologies such as Natural Language Processing, Machine Learning, Intelligent Document Processing, Deep Learning, and Generative AI. As you learn, understand how each of these applies to AML, financial crime, and risk management work. For example, consider Natural Language Processing, where software reads, processes, and comprehends both structured (database) and unstructured (news articles) text. Then think about how reading and understanding texts influence your compliance or risk management role. We also recommend compliance and risk professionals begin meeting with AI product providers. Listen, ask questions, observe demonstrations, and run short trials. We believe that leaving AI solely to the IT department or the AML technology team is not the right approach for executives and leaders. They need to understand AI just as well as other critical components of their programs.
Finally, any thoughts you’d like to leave with our readers on the future of AI in banking and how institutions can prepare to stay ahead?
I would reiterate that all AML, risk management, and compliance leaders must learn and understand AI as well as they understand every other part of their programs. For many of us in the industry for over 20 years, software and technology were seen as essential, but often as supporting elements of our programs. In many cases, oversight of the technology was delegated to subordinates or even the IT department. That approach won’t suffice with AI. AI will be so embedded in every aspect of AML, risk management, and compliance that I can’t imagine a successful executive who isn’t fluent in the language and technical details of AI. If you’re not, start reading and engaging in conversations on the topic as much as you can.
Stay Ahead of the Financial Curve with Our Latest Fintech News Updates!
