What You Need to Know
- Key takeaway #1Many members of the Senate Banking Committee see a clear need for regulation of AI in financial services.
- Key takeaway #2There is also a desire to strike the right regulatory balance to stay competitive with other countries using AI technology.
- Key takeaway #3Lawmakers discussed the use of AI in credit underwriting and suggested it might be a good idea to require companies to give their consumers or the government transparency into decision-making and to explain how AI tools are used in that process.
On September 20, 2023, the United States Senate Committee on Banking, Housing, and Urban Affairs held a hearing on “Artificial Intelligence in Financial Services.” The purpose of this hearing was to discuss the current and future applications of artificial intelligence (AI) in the financial services industry, aiming to assess the potential risks and benefits of AI in areas such as credit underwriting, algorithmic trading, fraud prevention, and consumer lending.
Summary of the Hearing:
Senator Sherrod Brown (D-OH), Chairman of the Committee, opened the hearing by acknowledging the positive impact of technology on the financial system, such as easier access to money through ATMs and the ability to check bank balances on smartphones. However, he also highlighted the negative consequences, including automation leading to high-frequency trading and the flash crash, as well as fraud and scams on peer-to-peer payment platforms. Senator Brown emphasized the need to assess the implications of AI on the financial system and ensure that it is used to benefit American consumers and savers, rather than exploit them. He expressed concerns about AI being used to rig the system for the benefit of Wall Street and Silicon Valley, and called for policies and regulations to protect consumers and promote a fair and transparent economy. He also raised concerns about the potential for AI to perpetuate biases and discrimination in consumer lending markets.
Senator Mike Rounds (R-SD), filling in for Tim Scott, the Ranking Member of the Committee, acknowledged the importance of the topic and the bipartisan effort to understand AI. He highlighted the effective use of AI in the financial services industry, particularly in fraud prevention and accurate forecasting. He emphasized the need for transparency and explainability in decision-making, especially in credit underwriting, and the importance of protecting data and investing in cyber infrastructure. Senator Rounds expressed the belief that Congress should take a pro-innovation stance to attract talent and compete with global competitors like China. He cautioned against overregulation that could hinder progress and mentioned the need to shape AI in a way that reflects important values.
The hearing then proceeded with the following witnesses:
- Melissa Koide, CEO of FinRegLab and former Deputy Assistant Secretary for Consumer Policy
Ms. Koide discussed the research conducted by FinRegLab to develop fair and transparent machine learning models for credit underwriting. She highlighted the potential benefits of machine learning in improving credit underwriting but also raised concerns about bias and discrimination. Ms. Koide advocated for greater transparency and explainability in credit underwriting processes to ensure fairness and accountability.
- Professor Michael Wellman, Division Chair of Computer Science and Engineering at the University of Michigan
Professor Wellman discussed the promises and risks of AI in financial markets. He raised concerns about algorithmic trading, market manipulation, and the concentration of information and ownership in AI-powered trading. Professor Wellman emphasized the need to reconsider market designs and governance mechanisms to address these concerns.
- Daniel Gorfine, founder and CEO of Gattaca Horizons and former Chief Innovation Officer of the Commodities Futures Trading Commission.
Mr. Gorfine emphasized the competitive advantages and opportunities presented by AI in financial services. He recommended a thoughtful policy approach to foster responsible AI adoption while mitigating risks. Mr. Gorfine highlighted the importance of innovation, clarity in risk management, modernizing data privacy frameworks, and collaboration between the public and private sectors to combat fraud and scams.
Credit Decisions and Fair Lending
The hearing explored the issue of bias in AI algorithms for credit decisions and fair lending, particularly in the context of home appraisals and mortgage lending. The discriminatory impact of algorithms on minority borrowers was raised by multiple Senators, who cited studies that showed disparities in loan approvals based on race.
Senator Brown stated that AI should be used to benefit consumers and not exploit them. He highlighted the potential for AI to exacerbate biases and disparities if it is encoded with underlying data that is biased. Senator Brown also emphasized the need for transparency and accountability in AI decision-making, particularly in consumer lending markets where AI models can automate and amplify biases that exclude black and Latino Americans. Senator Elizabeth Warren (D-MA) also discussed the importance of preventing discrimination in AI-driven lending decisions and highlighted the existing legal authorities that apply to the use of innovative technologies. She emphasized that lenders would remain legally responsible for outcomes that violate such laws, regardless of whether the discrimination is caused by AI. Here, Ms. Koide agreed that the Consumer Financial Protection Bureau (CFPB) has the authority to address discriminatory behaviors in AI-driven lending.
Senator Bob Menendez (D-NJ) expressed similar concerns about AI exacerbating existing disparities in access to financial services and asked how to address issues of accountability and oversight. Ms. Koide emphasized the importance of using data that is representative of full populations and monitoring AI models to make sure they do not discriminate. She highlighted the need for fair lending laws and automated techniques to identify less discriminatory model alternatives. Ms. Koide also noted the distinction between supervised machine learning algorithms and large language models, which require further attention.
Senator Chris Van Hollen (D-MD) raised the issue of AI manipulation in financial markets and questioned the potential liability of individuals who program AI systems to maximize profit without intending to manipulate prices. Responding, Professor Wellman explained that AI systems can independently learn to manipulate benchmarks to maximize profit and that the current laws built on human intent may not clearly address this issue. Additionally, Mr. Gorfine suggested that proper model building techniques should include controls and governance to prevent manipulation and that financial market regulators should upgrade their surveillance tools to detect new forms of manipulation.
Steps to Provide for an AI Workforce of Tomorrow
The hearing underscored the need for balanced and principled regulation of AI, emphasizing the importance of avoiding overly broad rules that could hinder small firms and community banks from adopting new technologies.
Senator Brown expressed concerns about big corporations potentially using AI in concerning ways, suggesting that their version of efficiency often leads to lower wages, fewer jobs, and less economic security for workers. Senator Brown also raised issues with the use of AI in financial services, particularly in consumer lending markets where AI models can perpetuate biases and exclude certain groups. He mentioned the rise of deep fake AI technologies used by fraudsters to deceive American savers and called for rigorous testing and evaluation of AI models to ensure they do not cause harm.
The hearing highlighted the need for policies and regulations to ensure that AI is used responsibly and for the benefit of consumers and society. Witnesses and senators emphasized the importance of transparency, accountability, and finding the right balance between regulation and innovation.
Mr. Gorfine discussed the need for a national framework to address risks related to data control, privacy, and use in AI. He cautioned against overregulation, which could hinder innovation and development, especially for smaller players in the industry. Mr. Gorfine also noted the importance of finding a balance between intervention and for technological advancements.
Mr. Wellman acknowledged the potential for AI to be used in fraud and scams, ranging from misinformation to new forms of fraud. He suggested leveraging AI to combat fraud, but also called for further study on the potential consequences and loopholes that may arise in an arms race between AI fraud and regulation.
AI Contributing to a Rise in Fraud in the Financial System
During the hearing, Senator Brown expressed concern about the potential for AI to compound the problem of fraud in the financial system. Senator Menendez noted that consumers are increasingly falling victim to sophisticated AI scams, including deepfakes and voice impersonation. He emphasized the need for financial institutions to take steps to minimize the risk of AI-powered scams targeting their customers.
Mr. Gorfine emphasized the importance of financial institutions investing in AI-related technologies to detect and prevent fraud. He also highlighted the need for collaboration between the public and private sectors in sharing information and threats. Ms. Koide discussed the potential of AI and advanced analytics to address fraud, but acknowledged that more research is needed in this area.
On the potential for AI to manipulate financial markets, Professor Wellman highlighted the use of reinforcement learning techniques by AI systems to maximize profits, which could potentially lead to market manipulation. He emphasized the need for regulators to upgrade market surveillance tools to detect and prevent these new forms of manipulation. Mr. Wellman also explained that existing legal and regulatory frameworks are not adequately covering new forms of fraud facilitated by AI.
This hearing comes after a series of hearings throughout the Senate Committees, highlighting another aspect of AI regulation. Here, the Committee’s focus overlaps with those heard across the U.S. Congress and underscores the need for balanced AI regulation that benefits the American people as well as allows for the advancement of technology across our global landscape.
Crowell & Moring, LLP will continue to monitor congressional and executive branch efforts to regulate AI. Our lawyers and public policy professionals are available to advise any clients who want to play an active role in the policy debates taking place right now or who are seeking to navigate AI-related concerns in financial services, as well as intellectual property, privacy, health care, and other areas.
For further information, please contact:
Aaron C. Cummings, Partner, Crowell & Moring
Note: Our lawyers leveraged AI in creating this client alert, including using a transcript summary created by generative AI. As we explore the potential of generative AI in the legal space, it is our intention and our practice to be transparent with our readers and to showcase the results we are achieving using generative AI with publicly available resources. Crowell’s AI group is comprised of lawyers and professionals across our global offices, including from Crowell & Moring International (CMI), our international public policy entity, with decades of sector-specific experience. We intend to lead by example in our own responsible use of AI, as it pertains to both the risks and benefits. Should you have questions about the use of generative AI in the legal sector or Crowell’s use of AI, please contact firstname.lastname@example.org.