Overview
- The Office of the Comptroller of the Currency issued the framework on Wednesday, Nov. 4, 2026, to guide examiners on banks' use of artificial intelligence and agentic systems.
- The document is nonbinding and technology-neutral, does not create new enforceable rules, and does not require banks to set a separate 'Super Intelligence' risk appetite.
- Examiners are instructed to assess AI by the likely consequences of failures rather than by model sophistication and to apply four illustrative risk tiers that put fully autonomous payment agents and real-time credit decision systems in the highest-risk category.
- To curb operational threats from external or third-party agents the framework recommends concrete controls including identity verification, delegated permissions, transaction limits, continuous monitoring, the ability to suspend access, circuit breakers, rate limits, emergency shutdowns, independent testing, and human approval for consequential actions.
- The guidance is expected to shape examiner focus and prompt banks that reported unauthorized third-party agents in late October to tighten controls, which could limit how customer or fintech AI assistants interact with accounts while the OCC plans to update the framework as technology evolves.