What changed. What you do next.
Short operating notes on AI supervision. Instruments, dates, and what a programme in build should do with them.
All notes
Filter by market or topic. Each note ends with an operating implication.
-
Model risk: convergence on mechanism, divergence on boundary
The commercial case is not that every supervisor invented a new AI statute. It is that the mechanisms rhyme and the boundaries do not.
-
A layered framework, not a monolith
HKMA circulars since 2019 look like fragmentation until you read them as layers. Packs encode the layers.
-
OSFI E-23 brings Canadian insurers into OSFI's enterprise-wide model risk regime for the first time
E-23 is not only a banking story. Insurers inherit OSFI model-risk discipline for pricing and claims AI, with a hard effective date of 1 May 2027.
-
What SR 26-2 actually changed
SR 26-2, issued 17 April 2026, replaces SR 11-7 and SR 21-8 with risk-based cadence and a tighter model definition that fits AI and ML.
-
Singapore's proposed AI risk management guidelines: what a twelve-month transition would mean
A twelve-month transition is long for a slide deck and short for a control plane. The clock has not started. MAS is still reviewing feedback.
No notes match those filters.
Map a note to a system you run.
Request a briefing. Forty-five minutes on that system and your jurisdiction. No slides.
Request a briefing