AI agents are already doing more of the work in disputes. From the moment a customer flags a charge to the final decision on the case, agents are handling intake, pulling transaction history, assessing evidence, and in some institutions making the call. For disputes teams, this raises questions that did not exist two years ago.

When an agent makes the wrong decision, who owns it? How do you walk a regulator through an automated process step by step when the steps are not always visible? And as institutions build stronger evidence trails to support agentic decisioning, how does that same infrastructure become one of the best tools for catching cases where the dispute itself is the attack?

First party fraud through the disputes channel is growing. Customers who are genuine account holders, not external bad actors, file fraudulent claims knowing that automated systems are faster and less scrutinizing than human reviewers. The same agentic infrastructure that speeds resolution can, with the right design, detect the patterns that expose those cases.

This roundtable brings together practitioners working through these questions in live programs. We will focus on what is working today, where the accountability gaps are, what a compliant and auditable agentic disputes process actually looks like in practice, and what needs to happen next for institutions to trust agents with more of this workload.

This roundtable is designed for senior fraud leaders responsible for operations and the technology/product side of fraud prevention at larger banks and credit unions.