OCC Bulletin 2026-54: First AI Supervision Framework for US Bank Examiners

The OCC (US regulator of national banks) published its first AI supervision framework. Examiners classify AI use cases based on three factors: (1) potential consequences of failure, (2) the shift of agency from humans to systems, and (3) the degree of human oversight — not technical sophistication. An official example: a simple AI agent authorized to initiate transactions could be riskier than a sophisticated model used for internal purposes. The framework sets no new supervisory expectations and banks are not required to maintain a separate AI risk appetite; the focus is on consequences and control intensity rising with potential impa