Federal Reserve Gov. Michael Barr said that while generative artificial intelligence holds great promise to boost efficiencies in a range of banking functions, important and as-yet-unresolved questions of liability in the event of an AI hallucination or AI-enabled security breach should spur banks to be diligent in their adoption of the technology.
Speaking at a fintech conference at the Federal Reserve Bank of San Francisco Friday morning, Barr said that generative AI poses a unique challenge for banking. The potential for AI to speed up and/or enhance many of the data-intensive choices banks make — such as underwriting a loan or aiding a customer — is immense, Barr said, but because banking is such a highly regulated and risk-sensitive business, banks to date have largely been unable to harness the technology to its fullest potential. That will inevitably change, Barr said, and banks should consider how to manage the inherent risks of AI before it becomes a competitive necessity.
"To get prepared for this moment, bank risk managers and regulators should become familiar with gen AI trends and monitor developments outside the bank perimeter so that they are not caught off guard as this technology quickly enters the banking system," Barr said.
Barr — who served as Federal Reserve vice chair for supervision until late February — said that the relationship between banks and fintechs with regard to generative AI has been both competitive and symbiotic.
Banks are large, incumbent institutions with large troves of customer data that can be better leveraged by AI, but often they silo information in firewalled departments for data security or organizational reasons. Digital-native fintechs, by contrast, have few of those organizational hurdles, but lack the scale and quality of customer data for AI to reach its peak potential. That has led many banks to partner with generative AI fintechs in a mutually beneficial arrangement, but Barr said that banks should be careful not to offer over their data without fully thinking through how that data could lead to liabilities down the road.
"To the extent banks are using gen AI or offering gen AI products and services, they have the responsibility to manage their risk, and should use their relationships to incentivize good risk management practices for fintechs," Barr said. "This means choosing fintech partners that provide transparency and clarity regarding the development of AI tools and have demonstrated appropriate control in deployment."
Barr added that reaching those agreements isn't always straightforward because banks will naturally want to know everything they can about an AI model that their customer data will be fed into, but fintechs are naturally wary of giving away too much of their competitive edge.
"There's necessary tension here, as banks must understand the tools offered by their fintech partners for their own risk management, while fintechs may not want to share details they hold close — their secret sauce," Barr said. "With respect to gen AI, it is important for fintechs and banks to tackle questions like who owns the customer data, and potential conflicts that may arise if a bank's customer data are fed back into a fintech's Gen AI model."
Regulators have a role to play in this dynamic as well, Barr said. Regulators should be proactive in reviewing their standards for AI model risk management, and engage with the industry and other relevant stakeholders with curiosity as they conduct that review. Regulators should be looking for ways to adopt AI for their own purposes as well, he said, both to "remain in touch in the changing world and make reasoned judgments" about how generative AI can and should be used in banking.
"These changes will require broad-based curiosity from regulators, fintechs and banks — combined with education and investment — to create a culture of awareness on the opportunity and risks of the technology," Barr said. "Equally as important is leadership, to establish appropriate governance over AI and provide appropriate direction on priorities."
