What does BCBS Digitalisation of finance report (May 2024) say about AI in banking?

What are the key points of BCBS Digitalisation of finance report (May 2024)?

• Technologies assessed: application programming interfaces, artificial intelligence and machine learning, distributed ledger technology, and cloud computing, plus new technology-enabled suppliers.
• Lists AI/ML use cases observed at banks: credit underwriting, trading, pricing, regulatory capital and liquidity planning, fraud detection, AML/CFT, chatbots, and marketing.
• Finds banks' use of generative AI 'remains limited at present, especially for customer-facing services and higher-risk activities', with pilots in lower-risk internal tasks.
• Generative AI risk list: model risk (reasoning errors, hallucinations, inconsistent outputs), explainability, data governance and bias, governance and accountability, and third-party risk including unclear allocation of liability with vendors.
• Financial-stability angle: broad GenAI adoption may increase interconnectivity, amplify procyclical behaviour, and raise third-party concentration risk, 'particularly where the model vendors are also providing cloud services to banks'.
• Table 2 catalogues bank mitigants for GenAI: firm-wide AI policies, human review of outputs, employee training, enhanced validation even for lower-risk uses, and formal approval by IT and model-risk functions.
• Eight implications for banks and supervisors across macro-structural elements, specific digitalisation themes, and capacity building and coordination.

What did BCBS Digitalisation of finance report (May 2024) change for banks?

This is the Committee's most detailed statement on AI and the first to address generative AI. It confirmed that Basel expects GenAI to be governed inside existing model-risk frameworks with graduated, risk-based controls, and it named vendor-plus-cloud concentration as a systemic concern, a theme later carried into the December 2025 third-party principles and the FSB's 2025–26 AI work.