US Bank Regulator Issues First AI Framework as Banks Battle Rogue Agents

The US Office of the Comptroller of the Currency has issued its first supervisory framework for AI in banking, directing examiners to focus on the consequences of errors and the controls around AI systems. It comes amid reports of unauthorised third-party agents acting inside bank accounts, and addresses how customers might use financial assistants without giving them excessive access.

The 22-page framework sets out four illustrative risk tiers, placing autonomous payment agents and real-time credit decisions at the highest level, while basic internal productivity tools sit at the lowest. Examiners may consider whether actions can be reversed and how much human control remains, with safeguards such as transaction limits, monitoring, testing and emergency shutdowns. The OCC says the guidance creates no new enforceable requirements and will be updated as the technology develops.