<- Take me back to my Risk x Corporate Banking (SG) overview
Executive Summary
Risk functions at Singapore corporate banks operate MAS Notice 637 as the framework for capital adequacy risk: capital ratio measurement, capital adequacy stress testing, capital instrument risk classification, and capital planning at the group level. Across the two findings in this cell, an AI model fabricated a parallel holding-company notice and misrepresented the meaning of yellow highlighting in MAS amendment PDFs. For risk functions, both failures translate into capital adequacy risk positions captured against the wrong regulatory basis.
How AI gets this regulation wrong
Both findings are inference drift. The AI produced confident answers on instrument identification and on amendment-convention reading; both answers are contradicted by MAS's own published text. Neither output telegraphs to the risk function that the underlying claim is unverified.
| AI's Failure Mode | Count | Affected findings |
|---|---|---|
| Exposed Fabrication | 2 | Finding#1 · Finding#2 |
What that means for your practice
For risk teams at corporate banks, both findings load onto capital adequacy risk management integrity: the bank's capital risk position is recorded, measured, and reported against the wrong regulatory basis if AI-derived content is used without verification. The exposure is concentrated in risk policy documents, capital adequacy risk-appetite statements, and the regulatory inputs feeding into capital planning.
| Risk Impact | Count | Affected findings |
|---|---|---|
| Regulatory enforcement / professional liability exposure | 2 | Finding#1 · Finding#2 |
When this affects Risk teams at Corporate Banking firms
Risk functions at corporate banks apply Notice 637 across capital ratio risk measurement, capital adequacy stress testing, capital instrument risk classification, group-level capital planning (particularly where the corporate bank operates under a Singapore financial holding company), and risk reporting to the board's risk committee.
The two findings map onto AI-assisted research patterns common in risk work. First, what regulatory instrument is the source of capital adequacy obligation at the holding company level. The AI's fabricated 'Notice FHC-N637' produces a risk policy citation that does not exist. Second, when reviewing the 2025 amendment for changes to capital risk treatment, how to read yellow-highlighted passages. The AI's wrong characterisation causes the risk function to capture editorial annotation as a live amendment to the framework.
The findings at a glance
The table below lists each finding from the AI testing on MAS Notice 637 in this cell, showing the topic, the AI's failure mode, and the citation identifier.
Aggregate impact
The two findings describe a generation pattern that risk teams should expect on AI-assisted Singapore prudential research. The model commits to fabricated regulatory instruments and to wrong convention characterisations on question types that are foundational to risk policy documentation.
For capital adequacy risk management, this matters because the failures feed directly into the records the function maintains: capital adequacy risk policies, risk-appetite statements on capital ratios, internal capital adequacy assessment process documentation, and capital planning models. Each of these records cites MAS instruments and references specific amendment-related changes; AI-derived citations and amendment characterisations entering these records embed regulatory-basis errors into the bank's risk infrastructure.
What your team should do
Risk functions at corporate banks should treat AI tools as a research-prompt generator on Notice 637 work, with a mandatory verification step against MAS's published text before AI output enters risk policy, risk-appetite, or risk-reporting documentation.
Practical safeguards: (a) every MAS instrument citation entering a risk policy or risk-appetite document must be verified against the MAS publications portal. (b) When reviewing the 2025 amendment for changes to capital adequacy risk treatment, pull paragraph 3 of the amendment to confirm the reading convention before logging changes. (c) Anchor group-perimeter capital risk positions on Notice 637 paragraph 11.2.2, not on AI-supplied parallel-instrument references. (d) Build into the second-line risk function's review of first-line capital adequacy work a specific verification step for AI-derived instrument citations and amendment characterisations.
Where AI tools support risk work on this notice: outlining risk policy structures, identifying which Parts of Notice 637 are relevant to a particular risk measurement question, and producing first-draft summaries of regulatory architecture for review against source.
How RLB Can Help
RegLeg's published Hallucination Research is available as a free pre-flight check for Singapore banking practitioners working across MAS-supervised entities. Before relying on AI-assisted output for regulatory interpretation, compliance advice, or capital-instrument structuring, practitioners can consult the research to identify where AI tools have demonstrably mis-stated the rules: invented instruments, misread editorial conventions, outdated paragraphs presented as current. The research covers specific MAS instruments and surfaces the exact questions where AI tools have failed, making it a practical reference rather than a general caution.
For firms where multiple teams are working the same regulatory portfolio, RegLeg offers bespoke deep-dives into individual MAS instruments. These engagements go beyond the published findings to examine the full pattern of AI failure modes relevant to the instrument: the question types, the failure mechanisms, and the risk implications for compliance, risk, treasury, legal, and reporting work. The output is designed to be shared across functions and used as a durable reference, reducing duplicated due-diligence effort and creating a consistent internal standard for AI-assisted regulatory work.
RegLeg also develops training and CPD-aligned content for Singapore banking teams. The material translates the failure-mode catalogue into practical guidance on the classes of error practitioners should watch for: confabulated cross-references, version confusion between superseded and current instruments, jurisdiction bleed between superficially similar regimes, and inference-driven elaboration that overstates what an instrument actually requires. Separately, RegLeg offers a confidential review of a firm's existing AI-use policy against the failure-mode catalogue, identifying gaps between the policy's assumptions and the documented evidence of how AI tools perform on Singapore prudential questions in practice.