The US Office of the Comptroller of the Currency issued a supervisory framework on November 4, 2026, to help examiners assess banks’ use of advanced artificial intelligence systems described by the agency as super intelligence and agentic super intelligence. The guidance addresses governance, oversight and risks associated with these systems, while expressly stating that it does not impose new regulatory requirements on banks.
In OCC Bulletin 2026-54, the regulator said its examiners should tailor reviews to the significance and potential consequences of each bank’s AI use case. The framework is intended as a reference for supervisory staff, not a separate set of enforceable standards or an instruction for banks to adopt the document as their own risk-management policy.
Examiners can use existing risk-management categories
The OCC said super intelligence should not be treated as an entirely separate bank-risk category. Examiners may use the guide to identify AI-specific considerations within existing areas such as model risk, compliance, technology, governance and operational resilience.
The regulator also stated that banks are not required to create a distinct risk appetite for super intelligence. This is a consequential limitation: the November 4 release does not establish a new AI licensing regime, prohibit financial institutions from developing agentic tools or order banks to implement a particular model architecture.
Instead, the OCC wants supervisory work to reflect how much autonomy an AI system possesses, the potential consequences of mistakes and the level of human control over its actions. The framework may become more relevant when an agent can initiate transactions, change records or execute a multi-step workflow rather than merely summarize documents.
Human oversight and possible consequences guide scrutiny
The bulletin notes that the intensity of controls may increase as the potential impact of an agent’s failure or misuse grows. A system that drafts a nonbinding internal memo poses different risks from one permitted to move funds or alter customer information.
For community banks, the OCC instructs examiners to consider institutional size, complexity and risk profile. Its approach is intended to be proportionate rather than a uniform requirement to deploy complex controls for every AI-assisted task.
The agency expects the framework to be updated as technologies, use cases and supervisory understanding evolve. That acknowledgement is significant because advanced AI deployment models are changing quickly, and controls suitable for one category of application may not transfer neatly to another.
What the framework means for digital-asset businesses
The new guide concerns bank supervision broadly and is not a crypto-specific rule. It could nevertheless be relevant when federally supervised banks evaluate AI for payment operations, compliance, custody administration or blockchain-related services.
There are clear limits to that inference. The OCC bulletin does not authorize cryptocurrency activities that would otherwise be restricted, expand a bank’s legal powers or resolve separate obligations relating to digital assets. Financial institutions must still comply with their applicable banking, cybersecurity, data-protection and financial-crime requirements.
BlockBriefing previously covered a proposed SEC crypto custody framework, which involves a separate regulator and policy question. Those developments should not be combined into a claim that the OCC has issued comprehensive new crypto AI regulations.
Investors and customers should also distinguish supervisory guidance for examiners from enforceable obligations imposed directly on banks. The OCC’s accompanying official release emphasizes a technology-neutral approach to safety and soundness.
The next step is how examiners apply the framework in ordinary supervisory activity and whether the agency revises it after receiving feedback. The immediate development is publication of examiner guidance, not a new statutory or regulatory AI mandate.
