Financial Services Innovate to Combat Fraud and Enhance Trust

In the rapidly evolving realm of financial services, innovation is crucial for addressing fraud and risk management. A recent discussion led by Matthew Pearce, the vice president of fraud risk management and dispute operations at i2c, emphasized the pressing need for balance between speed, security, and customer experience. The conversation, held during the B2B Payments 2025 event, highlighted how milliseconds can significantly impact transactions and relationships in digital payments.

The challenge lies in fraud detection systems that can alienate legitimate customers if they lag, while overly aggressive filters can trigger false declines, damaging trust and revenue. Pearce articulated that striking a balance is essential, noting three key metrics that i2c closely monitors: fraud loss ratio, fraud decline rate, and false positive rate. These indicators serve as benchmarks for the delicate equilibrium between vigilance against fraud and maintaining usability.

Innovative Solutions Through AI

Fortunately, advancements in artificial intelligence (AI) are fostering a more promising future for fraud management. As generative and predictive AI become integral to financial operations, they demonstrate that performance and accountability can coexist. Pearce remarked that leading institutions are continuously tuning their models to measure performance across multiple dimensions, ensuring they adapt without compromising core values.

He referred to this concept as “agility without volatility,” which encapsulates the need for systems that evolve quickly enough to counter fraudsters while remaining stable for existing portfolios. “Agility without volatility is the new definition of resilience,” Pearce stated, emphasizing that adaptability is just as important as accuracy in this dynamic landscape.

Building Trust Through Explainability

While AI is becoming indispensable in financial operations, it raises critical questions regarding trust and accountability. Regulators are increasingly scrutinizing “black box” decision systems, demanding transparency in areas such as credit scoring and fraud detection. Pearce emphasized that these requirements should be viewed as fundamental design principles rather than mere compliance checkboxes.

At i2c, every AI model undergoes rigorous versioning, documentation, and fairness testing prior to deployment. This ensures that when regulators or clients inquire about decision-making processes, the company can provide a clear narrative detailing the data lineage, rationale, and governance path leading to the outcome. “Every outcome must be traceable, from the features and rules behind it to the business impact that it creates,” Pearce explained.

Data plays a vital role in achieving this “full story” explainability. According to Pearce, the effectiveness of AI in payments is directly tied to the quality of the data that fuels it. “We draw insights from a broad mix of transaction data, dispute outcomes, and behavioral patterns,” he added. Each dataset undergoes schema checks, drift tracking, and challenger testing before a model is deployed, ensuring its reliability.

The approach taken by i2c includes a federated learning model, which combines a local and global perspective to enhance predictive power without bias. Pearce noted that models learn from global trends while adapting to local conditions, allowing for performance accuracy without overfitting to any single data source. Another important aspect is the exclusion of personally identifiable information in training pipelines, which is tokenized or hashed at the architectural level. This method guarantees that models only access attributes relevant to prediction, thereby safeguarding customer privacy.

As regulatory bodies like the Federal Reserve and the Consumer Financial Protection Bureau refine their frameworks for algorithmic accountability, financial institutions will require systems that not only perform effectively but also provide clear insights into their operational integrity. Pearce asserted, “Transparency never becomes the cost of security.” He emphasized that protecting privacy begins at the forefront of operations.

From Pilot Programs to Proven Impact

Even the most advanced AI systems face challenges without a clear implementation pathway. For banks and FinTech companies, the primary issue often lies not in what to build, but in how to operationalize these innovations. Pearce outlined a disciplined 90-day cycle for effective AI adoption: defining scope and success criteria, followed by integration, configuration, and limited rollout.

He pointed out that the toughest barriers are frequently organizational rather than technical. Factors such as governance, approvals, data quality, and regulatory compliance often present more significant obstacles than coding challenges. The emphasis at i2c is on achieving “proof of impact” rather than merely a “proof of concept.” By focusing on results, i2c aims to position AI as a strategic asset rather than an experimental venture.

This distinction resonates with financial institutions seeking tangible returns on investment from their digital transformation efforts. The integration of i2c’s solutions is designed to be seamless, employing APIs that coexist with legacy systems and customer relationship management tools, which alleviates the resource burden on clients. Pearce remarked, “Client resources stay light,” allowing clients to retain data access and compliance oversight while the provider handles setup and governance.

The ongoing work at i2c hints at the future of FinTech, with intelligent systems that are not only faster and more adaptable but also more ethical and auditable. As the financial services industry continues to navigate the complexities of fraud management, the integration of advanced technologies and a commitment to transparency will be crucial in building trust and ensuring security for customers worldwide.