Skip to content

GRA · 金融体系风险

空间金融资本与危害、风险暴露和转型的交汇之处

AI 与模型风险

Financial institutions rely more and more on models they did not build and cannot fully see: credit scoring, fraud detection, catastrophe models, trading systems and general-purpose AI from a few providers. The Global Risks Alliance (GRA) examines how model failure becomes financial loss: validation gaps, third-party concentration, weak human oversight. It works on shared questions of method, never on any firm’s own models. Part of Financial-system risk.

02传导

模型失效如何转化为金融损失。

渠道冲击如何传导至金融
验证缺口A model performs outside the conditions it was tested on, and losses or mispricing follow before anyone notices
第三方集中度Many institutions rely on the same model or AI provider, so one fault or outage is correlated across firms
Herding在相似数据上训练的相似模型,对同一信号作出相同的反应
监督失灵在没有人能够理解、质疑或叫停的情况下,自动化输出即被付诸行动
Data drift气候、人口或市场变化使历史训练数据难以作为可靠依据

渠道描述的是机制,并非实测强度或概率。

03监管框架

Model risk rules changed in 2026.

工具状态,2026 年 10 月涵盖内容
US revised interagency guidance on model risk management (SR 26-2)Issued April 2026 by the Federal Reserve, OCC and FDIC; supersedes SR 11-7Principles-based, risk-tailored model risk management; generative and agentic AI are stated to be outside its scope
PRA SS1/23 模型风险管理原则自 2024 年 5 月起对获得内部模型批准的英国银行生效模型识别、治理、开发、验证和风险缓释措施
EU AI Act, as amended by the AI Omnibus信用评分高风险义务现自 2027 年 12 月 2 日起适用Risk management, data governance, human oversight and documentation for high-risk AI, including creditworthiness assessment
NIST 人工智能风险管理框架Voluntary, 2023Govern, map, measure and manage functions for AI risk
04Systemic angle

When many institutions use the same model.

模型风险 就成为稳定性问题。再保险公司之间通用的巨灾模型、贷款机构之间通用的信用评分以及交易台之间通用的市场数据源,都可能同时朝同一方向变动。GRA 在行业层面研究这种共模风险暴露:哪些模型族是共用的,其输入发生变化时会怎样,以及存在何种独立质疑机制。

Method guidance continues at AI and Financial Model Risk; AI oversight governance at the Global Risks Forum (GRF) AI oversight and accountability.

05问题

GRA 共同研究的问题。

  1. 01

    复合冲击下的验证

    How should models be tested against conditions outside their history, such as compound physical and operational events?

  2. 02

    提供商依赖

    How concentrated is the industry’s reliance on a small number of model and AI providers, and what fails together?

  3. 03

    人工监督

    What does meaningful human review look like when a decision relies on a model no one in the room built?

  4. 04

    巨灾模型不确定性

    对于金融读者,应如何呈现危害模型之间的差异,而不是将其平均掉?

06边界

面向行业的方法,而非对某家公司的评审。

  • GRA does not validate, approve or certify any institution’s model, and issues no model rating.
  • 其成果为共享的方法和证据,在评审和权利许可的范围内发布。
  • No member’s confidential model information is exchanged with another.
  • Supervisory expectations remain with supervisors; GRA’s cyber and operational work covers the related third-party questions.
07已关联

联合体内的同一主题。

GRA 平台

人工智能与金融模型风险

专业社区: 人工智能与金融模型风险.

科学

人工智能科学

来自全球风险与创新中心(Global Centre for Risk and Innovation,GCRI)的人工智能与自主 系统 证据: 人工智能与自主系统.

治理

人工智能治理

由全球风险论坛(Global Risks Forum,GRF)召集的治理对话: 人工智能与数字治理.

09问与答

问题与解答。

Will GRA review our institution’s models?
No. GRA does not perform members’ own risk-management work. It develops shared methods and evidence that any institution can examine and apply under its own governance.
Does GRA set standards for AI in finance?
不是。GRA 不制定标准,也不认证任何事项。任何成员都无须采用 GRA 发布的任何方法。
模型风险专家如何做出贡献?
Through professional participation and working groups, once their terms are adopted. Register interest at the membership application form.
报告贡献

从本页贡献证据。

Reports 是联合体的季刊。您对该主题的了解可以成为下一期中的证据、案例或更正,署名归您,并在属于您本人作品时以 CC BY 4.0 发布。

全球风险联盟(Global Risks Alliance)
全球风险联盟(Global Risks Alliance)

向联盟提出风险暴露或资本问题。

信息和咨询现已开放。每项咨询都会获得一个编号,并告知下一步;申请、付款、算力分配和注册只有在相应团队、条款和审查能力就绪时才会开放。