APRA CPS 230 came into force on 1 July 2025. It is, in our reading, the cleanest operational-risk frame for AI workloads currently published by any prudential regulator in any market.
Why we recommend it as a reference
CPS 230 frames AI workloads as operational dependencies - critical operations, service providers and material arrangements - rather than as a separate technological category. This is the right frame.
Most AI-specific regulation focuses on the model. CPS 230 focuses on the operation the model serves. The difference is significant: a perfectly safe model serving a poorly designed operation is still a hazard.