What's Happening?
On November 4, the Office of the Comptroller of the Currency (OCC) released a new supervisory framework detailing how examiners should assess banks' utilization of 'Super Intelligence' (SI), a term encompassing technologies previously known as artificial
intelligence. This framework applies to SI used across various banking functions, including credit decisioning, payments, fraud detection, customer service, operations, risk management, and market activities, with a specific focus on generative and agentic systems. The OCC clarifies that SI is not a distinct risk category; instead, its associated risks must be identified, measured, monitored, and controlled within banks' existing risk management frameworks. A core principle of the framework is proportionality, meaning the potential consequences of an SI system's failure or misuse should dictate the strength of its controls. The framework outlines six pillars and sixteen supervisory components, covering aspects like purpose, risk, control, oversight, resilience, and assurance. For consumer-facing applications, the OCC emphasizes that SI does not alter banks' existing legal obligations, highlighting potential compliance risks from inaccurate or misleading outputs.
Why It's Important?
This supervisory framework is crucial for the U.S. banking sector as it provides a standardized approach for managing the evolving risks associated with advanced AI technologies. By integrating SI risks into existing risk management frameworks, the OCC aims to prevent the creation of regulatory silos and ensure a consistent approach to oversight. The emphasis on proportionality means that banks will need to carefully evaluate the potential impact of their SI systems, leading to more robust controls for high-stakes applications like autonomous payment agents or real-time credit decisioning. This could influence how banks invest in and deploy SI, potentially slowing adoption in critical areas until robust controls are in place. Furthermore, the explicit mention of consumer-facing uses and the unaltered legal obligations underscore the OCC's commitment to consumer protection, which could lead to increased scrutiny of SI applications in customer service, marketing, and disclosures. Banks that fail to adequately manage these risks could face significant compliance penalties and reputational damage.
What's Next?
The OCC has indicated that this framework is a 'living document,' implying that it will be updated as SI technology, bank use cases, and supervisory understanding evolve. This suggests ongoing engagement between the OCC and the banking industry to refine regulatory expectations. Banks currently deploying or considering generative or agentic systems will likely use this framework as a roadmap to evaluate and adjust their existing controls and human oversight mechanisms. Future OCC examinations will increasingly focus on the specific risks posed by autonomous and rapidly evolving SI systems, building upon established risk-management principles. This could lead to a period of adjustment for financial institutions as they align their SI strategies with the OCC's guidelines, potentially involving investments in new risk assessment tools, enhanced data governance, and more rigorous testing protocols. The continuous evolution of the framework also means banks will need to stay agile and adapt their internal policies and procedures accordingly.
Beyond the Headlines
The OCC's framework for 'Super Intelligence' signals a broader regulatory trend towards integrating emerging technologies into existing oversight structures rather than creating entirely new ones. This approach reflects a pragmatic recognition that the fundamental principles of risk management remain relevant, even as the technologies themselves become more sophisticated. However, the framework also highlights the unique challenges posed by generative and agentic AI, such as 'hallucinations,' 'prompt injection,' and 'scope drift,' which could have profound implications for financial accuracy and consumer trust. The emphasis on human oversight and the potential for systems to take inappropriate actions even while operational raises ethical questions about accountability in an increasingly automated financial landscape. This development could also spur innovation in AI safety and explainability within the financial technology sector, as banks seek to demonstrate compliance and mitigate novel risks. Ultimately, the OCC's stance will shape the responsible development and deployment of advanced AI in U.S. banking for years to come.
