On May 13, 2022, NCUA examination chief Kelly Lay told the House Task Force on Artificial Intelligence that the NCUA was testing machine-learning models on quarterly Call Report data and NLP on unstructured exam documents, that AI and ML algorithms 'must be tested to prevent the intrusion of underlying historical bias,' and that Congress should restore the NCUA's authority to examine third-party technology vendors, which lapsed in 2002.
| Document | NCUA testimony to House AI Task Force (2022) — Written Testimony of Kelly Lay, Director of the Office of Examination and Insurance, before the House Financial Services Task Force on Artificial Intelligence |
| Issued by | National Credit Union Administration |
| Type | Speech |
| Status | Final |
| Published | May 13, 2022 |
| Applies to | Congressional testimony; no direct application to credit unions |
| Official source | ncua.gov ↗ |
| Use cases | Fair lending & discrimination · Third-party & vendor AI · AI governance (general) |
What are the key points of NCUA testimony to House AI Task Force (2022)?
- NCUA's own AI: ML clustering of credit unions into risk categories from Call Report data; NLP pilots on board minutes and audit reports
- MERIT examination platform launched in 2021 to modernize a 25-year-old system; goal of 'predominantly virtual' exams within five to ten years
- Warned that AI/ML must be tested to prevent historical bias producing discriminatory outcomes
- Two-thirds of credit unions hold under $100 million in assets with about seven employees, limiting in-house AI and vendor due diligence
- Requested restoration of third-party vendor examination authority, noting vendors had rejected security recommendations on cost grounds
- Companion March 2022 NCUA white paper on third-party vendor authority made the same request
What did NCUA testimony to House AI Task Force (2022) change for banks?
This is the NCUA's most detailed public statement on AI risk in credit unions before 2025. It established the two themes that still define the agency's position: bias testing under fair-lending law, and the vendor-oversight gap that leaves AI sourced from third parties outside examiner reach.
What did the NCUA say about AI bias?
That AI and ML algorithms must be tested to prevent the intrusion of underlying historical bias that could enable discriminatory practices, consistent with fair-lending obligations under ECOA.
Why does the NCUA keep asking for vendor examination authority?
Temporary Y2K-era authority to examine technology service providers expired in 2002. Without it, the NCUA cannot inspect the AI vendors most credit unions rely on; GAO has recommended restoring the authority since 2015 and again in 2025.
| Date | Document | Status |
|---|---|---|
| Feb 12, 2026 | Hauptman Senate testimony (Feb 2026) — Written Testimony of NCUA Chairman Kyle S. Hauptman before the U.S. Senate Committee on Banking, Housing, and Urban Affairs | Final |
| Jan 14, 2026 | NCUA Letter 26-CU-01 — NCUA's 2026 Supervisory Priorities | In force |
| Sep 1, 2025 | NCUA AI Compliance Plan (2025) — NCUA Artificial Intelligence Compliance Plan | Final |
| Aug 1, 2025 | NCUA Credit Union AI Resource Center — Artificial Intelligence (AI) — Credit Union AI Resource Center and AI Resources page | In force |
| Jul 24, 2025 | NCUA Board AI briefing (Jul 2025) — Board Briefing: Artificial Intelligence and Its Use Within the Credit Union Industry and the NCUA | Final |
| Sep 21, 2023 | NCUA Financial Innovation Rule (2023) — Financial Innovation: Loan Participations, Eligible Obligations, and Notes of Liquidating Credit Unions (final rule) | In force |
Follow every document these regulators publish
6 curated AI stories for banking executives · Every morning · Free
Subscribe to BankingNewsAI →