How Banks Can Stay on Top of Emerging A.I. Risks and Regulations

This article was expanded for additional publication by Alabama Bankers Association in Vol. 34, Issue 2 of Banking Traditions Magazine. 

The convergence of big data, computing power and generative A.I. algorithms is already having major impacts on the banking industry. Financial institutions can stay ahead of evolving regulations by reviewing their products and operations and implementing best practices now.

A.I. Is Changing the Way Banks Do Business

Artificial intelligence involves machine functions that emulate human process and communication. Learning systems improve performance over time using training and experiences. New regulations surrounding the banking industry and its use of these new technologies require fairness, reliability and transparency from institutions employing them. Banks should be aware of deceptive advertising, unfair competition, copyright and privacy issues, and false information that may result from the use of generative A.I.

Generative A.I. can easily output new data and “train” new data through generative adversarial networks (GANs). A large language model (LLM) analyzes human language patterns to easily predict words and review large quantities of unlabeled text to quickly sift through data. This type of A.I. can make several banking business functions more efficient, including:

• • Customer service
  • Fraud detection and prevention
  • Investing and portfolio services
  • Loan applications
  • Mortgage processing
  • Retirement planning

Recent developments in A.I. offer benefits to banks, but they’re not without risk. Risks to consumers can include poor data quality resulting from incomplete data or the release of unauthorized information, as well as a lack of transparency. The use of A.I. in financial services also may violate regulatory anti-discrimination obligations without human oversight or the option to override data. 

The Rise of A.I. Regulations

On March 31, 2021, the Department of the Treasury released a notice outlining the use of A.I. by financial institutions and the appropriate protocols for managing the technology to follow laws and regulations and protect consumers. These guidelines were the first in a series of regulations, followed by requirements for making credit decisions based on complex algorithms and quality control standards for automated valuation models, which were released in June 2023.

Top Tips for Banks Using A.I.

Banks can exercise the following options to protect consumer data and comply with federal regulations while realizing the opportunities that A.I. offers:

• • Conduct frequent risk assessments - Routine testing and monitoring of the choices A.I. tools are making gives your institution the chance to step in and further train the tools, as well as correct any issues. Focus on establishing protocols for data quality diligence, examining the mitigation framework, examining the pathways that direct the technologies’ decision-making, and assessing the output’s “explainability” – the ability to explain its decisions in a way that makes sense. You should also review the content of consumer notices of adverse decisions against applicable regulatory requirements for such notices.
  • Create an A.I. governance framework – Many of the risks that come with using A.I. involve a lack of human oversight. In addition to developing an A.I. incident response plan, it’s a good idea to update policies and procedures for every area of operations the use of A.I. could affect, including:
    • Business development
    • Compliance reporting
    • Consumer notices of adverse decisions
    • Data security
    • Management and staff training
    • Vendor management

Banks can find valuable resources on compliance requirements in the NIST Trustworthy & Responsible AI Resource Center, NIST AI Risk Management Framework, and Interagency Guidance on Third-Party Relationships: Risk Management.

A.I. and machine learning tools will continue to evolve, and so will the risks and regulations associated with them. A solid game plan can help banks reap the rewards of A.I. while managing any potential pitfalls. Please contact Drew Patty or any member of Phelps’ A.I. and Finance and Lending teams if you have questions or need compliance advice or guidance.