The latest compliance news: regulatory developments, punishments, guidance, permissions and authorisations for new product and service offerings.

Monetary Authority of Singapore
Along with a group of financial institutions and fintechs, the Monetary Authority of Singapore has published an industry white paper setting out how AI agents deployed in financial services can be kept within safe operating bounds as they take on more autonomous tasks.

The paper, entitled Safeguards for Agentic Finance at Runtime (SAFR), sets out an industry-developed framework for allowing AI agents to carry out financial tasks safely, securely and reliably, MAS said in a recent statement.

The framework was developed under MAS' BuildFin.ai initiative, which supports responsible development and deployment of AI in the financial sector.

The publication is the latest sign of how seriously Singapore's regulator and the Asian city-state's wealth and banking industry treat the shift from AI as a decision-support tool to AI as an active, autonomous participant in financial workflows. This change is particularly relevant for wealth managers and private banks, which handle sensitive client data and are subject to strict fiduciary and compliance obligations.

MAS said that as AI agents increasingly carry out tasks autonomously and at a speed beyond practical human intervention, financial institutions need real-time safeguards to ensure that agents' behaviour stays within the mandates, policies and risk boundaries that firms set for them. The SAFR framework proposes a set of governance checkpoints that verify and record an AI agent's proposed actions before it executes a task.

MAS said the framework builds on the AI risk management toolkit produced under MAS' Project Mindforge. However, it focuses specifically on how safeguards can be operationalised at the point where an agent acts, rather than only at the design or review stage.

Industry participants have already tested the framework across several use cases, MAS said. These include agent-assisted payments and treasury operations, where autonomous agents execute routine transactions within set mandates to cut operational friction; wealth management and advisory workflows, where AI agents review documents and produce structured assessments within narrowly-defined task boundaries to support faster, more consistent compliance review; and client engagement, where agents draft client insights and materials within approved content boundaries, freeing relationship managers and advisors to focus on higher-value client interaction.

MAS is inviting interested industry partners to join the BuildFin.ai work group to help shape future iterations of SAFR. The regulator's recently-announced Future of Finance Institute is expected to support adoption of the framework by facilitating industry pilots and sandbox experimentation, giving financial institutions a route to test and deploy SAFR-aligned solutions before wider rollout.

Apart from issues such as governance and safeguards in AI, another topic is the intersection of a client’s data privacy and AI – and how this is managed.