AITOPAI热报
USGuidelineAdoptedFinanceDeployers

OCC Risk-Tiering Framework for AI Agents at Banks

OCC 银行 AI 智能体风险分级框架
OCC·Published 2026-11-11·Checked by 管理员 on 2026-11-15

The OCC, the primary regulator of US national banks, issued a risk-tiering framework for assessing AI agents deployed by banks. The framework anchors its judgment on an agent's autonomy and permissions rather than model size: any AI agent capable of autonomously moving funds is placed in the highest risk tier, alongside real-time credit decisions and market access.

Source ↗Remind meUS · All

What it means for you

  • National banks deploying AI agents that can autonomously move funds face supervisory review at the highest risk tier, significantly raising compliance preparation costs
  • Human intervention mechanisms, audit trails, and compliance documentation become mandatory preconditions for launching payment-related AI agents
  • For agent developers, product design choices such as restricting fund-movement permissions directly determine the compliance barrier to entering the US banking system

    • What to do

    ComplianceInventory existing AI agents and identify any with autonomous fund-movement capability for top-tier classification
    EngineeringBuild human-in-the-loop mechanisms and full audit trails for payment AI agents
    LegalPrepare supervisory review compliance documentation in advance under top-tier standards
    ProductEvaluate vendor agents' fund-permission design as a procurement criterion

    Key dates

    2026-11-11Published

    Source documents

    Source ↗

    Coverage

    Stories in the feed
    Editorial summary, not legal advice; dates and status follow the original document. · Report a correction