Banks struggle to unlock AI’s full fraud-fighting potential
Chad Hetherington, global head of product at NICE Actimize, has spent years observing how artificial intelligence reshapes financial services, yet he cautions that its full potential often goes unrealized. Fraud losses remain on the rise, regulatory scrutiny intensifies, and many banks treat AI as a temporary experiment rather than a foundational tool. His solution centers on three principles: governance, transparency, and seamless integration into daily operations—or risk wasting millions on underperforming systems.
At NICE Actimize, Hetherington oversees enterprise-grade solutions for anti-money laundering, fraud detection, and compliance. The company’s cloud-based platform automates investigations, ensures regulatory compliance, and prioritizes customer experience. However, the greater obstacle lies not in developing the technology itself, but in ensuring it delivers measurable results. AI creates the greatest value when institutions understand where operational friction exists and apply intelligence to measurable business challenges, such as reducing fraud losses, accelerating investigations, improving customer onboarding, or strengthening regulatory compliance.
Successful implementations begin with specific business objectives. For example, a payment processor might use AI to reduce fraud-related chargebacks, while a bank could streamline customer onboarding. The difference lies in connecting AI directly to workflows rather than treating it as an isolated experiment. AI initiatives deliver lasting value only when they are embedded into enterprise workflows and measured against business and regulatory outcomes.
Why siloed data blocks AI’s full potential
Yet even with strong data and clear goals, many institutions falter at governance. Disconnected systems, fragmented data, and evolving regulations create barriers to safe, large-scale AI adoption. Financial firms often maintain separate tools for compliance, fraud detection, and customer data—each operating in isolation. Without a unified approach, AI cannot detect emerging risks or provide regulators with clear decision rationales.
“Customer information, fraud signals, transaction history and compliance data often reside in separate systems, preventing institutions from building a complete picture of customer behavior or emerging risk,” Hetherington says. “Modern AI platforms address this challenge by unifying intelligence across channels while preserving the governance, privacy and security controls required by financial institutions. The result is a more complete customer and risk view that improves operational efficiency, strengthens decision-making and enhances regulatory confidence.”
Trust requires more than functionality. Hetherington insists AI systems must be explainable, auditable, and subject to human oversight from the outset. “That insight drives quicker decision-making, tailored client engagement and streamlined operational processes. Organisations that successfully modernise customer engagement treat AI as a strategic business initiative rather than simply a technology deployment.”
Integration and leadership: The missing links
For fintech executives, the challenge extends beyond rapid AI adoption. Integration into the broader business model is essential, requiring leadership support, workforce training, and ongoing performance tracking. “Organisations that successfully modernise customer engagement treat AI as a strategic business initiative rather than simply a technology deployment. Leadership teams must establish clear objectives, align technology investments with measurable business outcomes and create accountability across the organisation. Equally important is investing in employees by providing the training, governance and operational processes that ensure AI becomes a trusted part of day-to-day decision-making.”
Hetherington’s stance is clear: AI must be reliable by design, not by accident. The company’s cloud platform unifies AI across financial crime domains, including anti-money laundering, enterprise fraud, payments surveillance, and case management. This integrated approach allows institutions to cross-reference transaction data, customer profiles, and regulatory signals in real time. For instance, a suspicious retail payment might trigger an automated AML check, while a fraud alert in a digital bank could prompt proactive customer notifications.
Beyond fraud and compliance, the platform enhances payments surveillance by monitoring transaction patterns for red flags like structuring, trade-based money laundering, or sanctions violations. Trade surveillance tools analyze cross-border flows to uncover irregularities, such as over-invoicing or misclassified goods, that might otherwise evade detection. These systems provide actionable context, such as linking payments to high-risk jurisdictions or previously flagged entities, which strengthens regulatory justifications.
Scaling AI requires unified governance and data
Scaling AI beyond pilot programs demands more than technical setup; it requires organizational alignment. Hetherington identifies three key components: trusted data, clearly defined business outcomes, and strong governance. Data governance ensures AI operates on accurate, consistent datasets. Many financial institutions struggle with fragmented data, where fraud teams use one system, compliance another, and customer service a third. NICE Actimize’s platform merges these sources into a single governed layer, ensuring AI decisions rely on up-to-date information.
Workflow integration is equally vital. Standalone AI tools, like isolated fraud detection modules, rarely sustain value. Instead, institutions must embed AI into existing processes, such as onboarding, transaction monitoring, or regulatory reporting. A bank using AI for automated KYC checks can cut manual reviews while maintaining compliance, but only if the system integrates tightly with identity verification workflows. Hetherington notes the most effective deployments treat AI as an extension of human decision-making, not a replacement. This means designing interfaces that allow overrides when necessary, while capturing those exceptions to refine models continuously.
Technology alone cannot drive transformation, people and processes must evolve together. Hetherington emphasizes that successful AI adoption depends on executive commitment, workforce training, and clear accountability. Leadership must define AI’s role in business strategy, allocate resources accordingly, and hold teams responsible for measurable outcomes. Without this support, AI initiatives often remain in pilot mode, lacking the scale needed to deliver impact. Training is equally critical; employees across fraud, compliance, and operations must understand how to use AI tools, interpret their outputs, and escalate exceptions.
