Artificial intelligence is increasingly embedded in the financial system’s decision-making architecture. It shapes how markets price risk, how credit is allocated, how institutions respond to stress, and how supervisors identify vulnerabilities. This Policy Letter organises the financial stability issues around five channels: shared models that could synchronize behavior, limited understanding and opacity, reliance on common providers, the dual use nature of AI in cybersecurity, and increasing autonomy that can blur accountability. It examines how these channels could play out in credit markets, market functioning, operational dependencies, and cyber risk, and sums up the challenge in three words: augmentation, speed, and explainability. Two cross-cutting principles follow for policy: proportionality, and the recognition that financial-sector resilience alone is insufficient. The impact of AI on financial stability is not predetermined; it depends on choices made by financial institutions, technology firms, and policymakers.
Policy Letter No. 122