Basel Committee connects AI risk with changes to bank disclosures

AI and interconnected financial risks

The Basel Committee’s 1 October account of its September meeting brings two developments together: the growing financial and operational footprint of AI, and the difficulty of comparing banks’ published risk information. Members discussed AI-related financing connections, leverage and shared dependencies, alongside possible vulnerabilities in critical financial functions.

A different format for disclosures

The committee approved a final standard for machine-readable Pillar 3 disclosures, with publication expected around the end of the year. It describes the prevalence of PDF-only disclosures as an obstacle to aggregating and comparing bank data. Approval at the meeting and publication of the final standard are distinct events.

What remains in progress

It also agreed to assess whether existing operational-loss categories adequately cover cyber risk and AI developments. Separately, additional guidance on interest-rate risk in the banking book is planned for consultation in November. These are work programmes and planned consultations, rather than evidence that a new requirement is already operating at every bank.

The documents to follow

For readers tracking financial institutions, the next useful documents are the published disclosure standard and the announced consultation. They will specify the reporting and supervisory changes more precisely than the meeting statement. The committee sets international standards; implementation depends on the relevant jurisdictions.

Continue with related research →