Business & Markets
OCC gives bank examiners an AI framework, with no new rules for lenders
The framework asks examiners to consider how much independence an AI system has and how closely people supervise it.

Executive summary
- Framework is for bank examiners: The Office of the Comptroller of the Currency issued guidance for examiners reviewing banks' use of artificial intelligence. Its bulletin uses the label "super intelligence" and says the term covers technologies previously encompassed by artificial intelligence.[1][2]
- OCC says it creates no new bank requirements: The OCC says the framework is an examiner resource, not a set of new supervisory expectations, enforceable standards or prescriptive requirements for banks.[1]
- Review focuses on potential consequences: Examiners should consider a system's materiality, autonomy, potential consequences and human oversight, rather than treat the technology as a separate risk category, the OCC said.[1]
The Office of the Comptroller of the Currency on Wednesday issued a framework to help examiners evaluate banks' use of artificial intelligence in ordinary business activities. The OCC bulletin uses the label "super intelligence" for technologies previously encompassed by artificial intelligence, citing an October executive order.[1][2]
The bulletin says the framework does not establish supervisory expectations, and banks are not expected to use it to assess their own systems. It also says the document creates no new regulatory expectations, enforceable standards or prescriptive requirements. Reuters described the framework as nonbinding guidance, while noting that banks typically comply closely with such supervisory guidance.[1][3]
Instead, the OCC says examiners may use the resource to tailor existing supervisory work to a bank's size, complexity and the materiality of a particular use case. The agency said it intends to update the framework as technology and supervisory understanding develop.[1][2]
The OCC says examiners should not treat "super intelligence" as a distinct risk category or require banks to set a separate risk appetite for it. The framework instead points supervisors toward the possible consequences of a system's failure or misuse, how much agency has been transferred to it, and the amount of human oversight.[1]
Under the OCC's example, a relatively simple agent authorized to initiate transactions could warrant more scrutiny than a more sophisticated model used internally. The framework says control intensity should rise with the potential consequences of a failure or misuse.[1][3]
The OCC supervises national banks and federal savings associations, among other institutions. The bulletin says examiners may apply the framework to community banks as well, tailoring reviews to each institution's risk profile and the nature of its systems. The agency said existing laws and regulations continue to apply.[1][2]
Reuters called the document the OCC's first guidance for how examiners should probe banks' AI use. That description refers to an examiner framework; the OCC's bulletin explicitly says it is not a new rule or an instruction for banks to create separate AI controls.[1][3]