Q&A

What is a typical document package requested by a bank from its AI vendor? 

(Similar questions)

• Can the group share their best practices in managing the risk of AI when it is an auxiliary part of broader vendor solution? 

• Most institutions will get AI through their core provider, fraud and AML vendors, or fintech partners rather than build it. What are institutions asking vendors about the AI embedded in their products (training data, model changes, explainability, contractual rights to information) and what answers are they actually getting?

Are AI and generative AI tools treated as “models” under your model risk management program  (inventoried, tiered, and validated) or are they managed under a separate framework? What has worked, and what has been hard to fit into the existing program?

Which AI training programs or certifications have members completed or are considering and what would you tell a colleague deciding whether one is worth the time and cost?

How are other banks using AI besides ChatGPT-type of questions and answers? Are your colleagues at your bank accepting information produced by AI?

Who owns AI governance at your financial institution? (See the polling results below)