Skip to content

GRA · Risiko sistem keuangan

Keuangan spasialTempat modal bertemu bahaya, keterpaparan, dan transisi

Risiko AI dan model

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.

02Transmisi

How model failure becomes financial loss.

SaluranBagaimana guncangan mencapai sektor keuangan
Validation gapsA model performs outside the conditions it was tested on, and losses or mispricing follow before anyone notices
Konsentrasi pihak ketigaBanyak lembaga bergantung pada model atau penyedia AI yang sama, sehingga satu kesalahan atau gangguan saling berkorelasi di berbagai perusahaan
HerdingSimilar models trained on similar data respond the same way to the same signal
Oversight failureAutomated outputs are acted on without a person able to understand, challenge or stop them
Pergeseran dataClimate, demographic or market shifts make historical training data a poor guide

Saluran menggambarkan mekanisme, bukan kekuatan atau probabilitas yang terukur.

03Kerangka pengawasan

Model risk rules changed in 2026.

InstrumenStatus, Oktober 2026Apa yang dicakupnya
US revised interagency guidance on model risk management (SR 26-2)Issued April 2026 by the Federal Reserve, OCC and FDIC; supersedes SR 11-7Manajemen risiko model yang berbasis prinsip dan disesuaikan dengan risiko; AI generatif dan agentik dinyatakan berada di luar ruang lingkupnya
PRA SS1/23 Model risk management principlesIn force for UK banks with internal model approval since May 2024Model identification, governance, development, validation and risk mitigants
EU AI Act, as amended by the AI OmnibusCredit-scoring high-risk obligations now apply from 2 December 2027Manajemen risiko, tata kelola data, pengawasan manusia , dan dokumentasi untuk AI berisiko tinggi, termasuk penilaian kelayakan kredit
Kerangka Manajemen Risiko AI NISTVoluntary, 2023Fungsi tata kelola, pemetaan, pengukuran, dan pengelolaan untuk risiko AI
04Systemic angle

Ketika banyak lembaga menggunakan model yang sama.

Model risk becomes a stability question when the same vendor model, data set or foundation model sits behind many decisions. Catastrophe models used across reinsurers, credit scores used across lenders and market-data feeds used across trading desks can all move in the same direction at once. GRA studies that common-mode exposure at sector level: which model families are shared, what happens when their inputs shift, and what independent challenge exists.

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

05Pertanyaan

Pertanyaan bersama yang ditangani GRA.

  1. 01

    Validasi di bawah guncangan majemuk

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

  2. 02

    Provider dependency

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

  3. 03

    Pengawasan manusia

    Seperti apa tinjauan manusia yang bermakna ketika suatu keputusan bergantung pada model yang tidak dibangun oleh seorang pun di ruangan itu?

  4. 04

    Catastrophe model uncertainty

    How should differences between hazard models be shown to financial readers, rather than averaged away?

06Batas wilayah

Methods for the industry, not reviews of a firm.

  • GRA does not validate, approve or certify any institution’s model, and issues no model rating.
  • Keluarannya berupa metode dan bukti bersama, yang diterbitkan sejauh tinjauan dan hak memungkinkan.
  • 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.
07Terhubung

Pokok bahasan yang sama di seluruh Konsorsium.

Platform GRA

Risiko AI dan Model Keuangan

Komunitas spesialis: Risiko AI dan Model Keuangan.

Sains

Sains AI

Bukti tentang Sistem AI dan otonom dari The Global Centre for Risk and Innovation (GCRI): AI dan Sistem Otonom.

Tata kelola

AI governance

Governance dialogue convened by The Global Risks Forum (GRF): AI and Digital Governance.

09Pertanyaan dan jawaban

Tanya jawab.

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?
No. GRA sets no standard and certifies nothing. No member is required to adopt any method GRA publishes.
Bagaimana spesialis risiko model dapat berkontribusi?
Through professional participation and working groups, once their terms are adopted. Register interest at the membership application form.
LaporanBerkontribusi

Kontribusikan bukti dari halaman ini.

Reports adalah publikasi triwulanan Konsorsium. Apa yang Anda ketahui tentang subjek ini dapat menjadi bukti, kasus, atau koreksi dalam edisi berikutnya, dengan kredit atas nama Anda dan dipublikasikan di bawah CC BY 4.0 jika merupakan karya Anda sendiri.

Global Risks Alliance
The Global Risks Alliance

Ajukan pertanyaan eksposur atau modal kepada Aliansi.

Informasi dan pertanyaan sudah terbuka hari ini. Sebuah pertanyaan menerima nomor referensi dan menyebutkan langkah berikutnya; lamaran, pembayaran, alokasi komputasi, dan pendaftaran hanya dibuka ketika tim, ketentuan, dan kapasitas tinjauannya tersedia.