Yields ads

Build your own model risk management platform, or buy a specialised one? This article puts both side by side. It breaks down the real costs of an in-house build, the years of specification work, the maintenance funded by one bank, the constant re-work as regulations change, and shows where a specialised platform changes the maths, backed by real examples from banks that have tried both.

The EU AI Act timeline has shifted, but the regulatory pressure remains. Prohibited practices and GPAI governance rules are already being enforced, with new deepfake labeling mandates taking effect as soon as August 2026. Understanding exactly what has been postponed, and what remains urgent, is essential for keeping compliance frameworks on track.

OSFI E-23 takes effect in Canada on May 1, 2027. It diverges sharply from the US approach under SR 26-2. E-23 applies a five-stage lifecycle to every model and brings generative and agentic AI fully in scope. SR 26-2 scales oversight to a model's material risk and keeps gen AI out. For institutions operating in both markets, the same model can fall under two different regimes. The full article breaks down where they diverge and how to manage both without duplicating work.

There are no free passes under the EU AI Act. Even minimal-risk AI systems must be thoroughly assessed and documented to prove their status. Proactive governance ensures your entire AI portfolio remains completely defensible against evolving regulatory demands.

SR 26-2 has officially rewritten the US model risk rulebook. Aligning your technology with this new reality ensures your institution moves smoothly toward risk-proportionate, dynamic orchestration.

SS1/23 came into force in May 2024. In 2026 the grace period is over, and the PRA has shifted from asking whether you have a policy to asking for evidence, lineage, and board-level accountability. This article sets out what that means in practice: an enterprise inventory that captures AI models and the material spreadsheets behind business decisions, conceptual soundness with no black boxes, and a named senior manager on the line under SM&CR. It also covers the April 2026 LIAF01/26 update, which brought stress-testing models under the same validation regime and tightened personal sign-off. A clear read on what the PRA expects today and what UK model risk teams need in place to meet it.