The National Association of Federally-insured Credit Unions is asked the U.S. Consumer Financial Protection Bureau to be more transparent about potential supervisory use of artificial intelligence (AI) and machine learning (ML).
In a letter to CFPB Director Rohit Chopra, Andrew Morris, NAFCU Senior Counsel for Research and Policy, asked the bureau to be more forthcoming about its intent to leverage AI within the supervisory prioritization process, especially in regard to its specific use.
Morris said the CFPB has focused on AI’s impact on consumers, but that it “is equally important that regulatory use of AI and LM follow similar standards of transparency and accountability.”
“Some important information regarding AI/ML projects comes not from the CFPB itself, but from vendors selected to perform data analysis contracts,” wrote Morris. “For example, a press release from one company describes a ‘Data Analytics Blank Purchase Agreement’ covering ‘machine learning, and natural language processing support.’ However, the CFPB has never disclosed where ML or natural language processing support may be applied.”
Morris said when it comes to data analytics, the CFPB’s contracting does not reveal whether awards cover the use of AI.
“For example, a contract that started on September 15, 2020 for complaint data analytics called for ‘applications to enhance CFPB capability to handle increasing volumes of complaints,’” the letter said. “The CFPB’s potential use of AI/ML techniques to assist supervisory prioritization should not be relegated to the fine print of federal contracts. Instead, NAFCU asks that the CFPB follow its own advice for financial institutions by ensuring that complex algorithms and models employed by the agency do not operate as black boxes.”
In his letter, Morris called on the CFPB to disclose “all AI/ML applications, services, and contracts that might bear upon the agency’s supervisory work or rulemaking activity.”