Artificial intelligence poses numerous benefits and risks to banks, said Acting Comptroller of the Currency Michael Hsu on June 16 during the American Bankers Association’s Risk and Compliance Conference.
Hsu called for banks to develop AI and tokenization in “tightly controlled stages where the risks can be identified, measured and managed at each stage,” with engagement by risk managers and compliance professionals. “The benefits of each are potentially quite significant, as are the risks — to consumers, to safety and soundness, and to financial stability,” he added.
To Hsu, alignment is the main challenge in safely developing AI. “Systems, which are generally based on neural networks, are not programmed explicitly like most software,” he said. “They require training, and their outputs are not predictable.”
Hsu said questions remain on who is responsible for the performance of AI when a bank uses a third-party implementor. AI systems also pose bias and discrimination challenges, he said. Generative artificial intelligence, which includes creating text or images by prompt, can enable misinformation.
“The ability of AI agents to mimic human communication and the low cost of scaling AI agents increase opportunities for fraud,” Hsu said. “The speed and sophistication of such developments warrant close monitoring and coordination.”
Hsu’s comments came as 70 percent of financial firms already use AI to detect fraud, predict the probabilities of loan default or late payments, personalize consumer recommendations and marketing and automate customer account openings.
Congress and other regulatory agencies are also evaluating the use of AI. During a May Congressional hearing on regulating artificial intelligence, OpenAI CEO Sam Altman agreed with a suggestion from Sen. Richard Blumenthal (D-Conn.) to create independent testing labs to inform customers on the trustworthiness of specific AI-generated content. During the hearing, New York University researcher Gary Marcus and Altman advocated Congress form a new agency to regulate AI, which would authorize licenses to AI providers.
In May, the Consumer Financial Protection Bureau warned banks that they could violate federal consumer financial protection laws if chatbots fail to provide accurate information to customers and don’t protect their privacy and data. In March 2021, the Federal Reserve Board, CFPB, FDIC, OCC and National Credit Union Administration issued a request for information on how financial institutions use AI.