AI Hallucination ResearchFindings by audienceSectorsInternational / MultilateralRetail BankingRiskDetail › Finding
Retail Banking × Risk — International / Multilateral · Last updated 11 Jun 2026 · Hallucination Register
Share / Print X LinkedIn Email

Distorted fast payment system count

RLB Citation ID: RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q010-Opus47
What the RLB Specialist Panel found
For Claude Opus 4.7 (web search on)
Question (paraphrased to protect IP)

A market briefing on the global fast payment system landscape needs CPMI data on how many domestic fast payment systems are currently operational globally, how many have already enabled cross-border payment exchanges, how many are planning cross-border linkages, and what proportion are operated by central banks versus private entities.

RLB's analysis

The model retrieved a 2025 monitoring-survey figure (57 systems) and presented it as the answer to a question about the global FPS universe. The November 2023 Tara Rice speech documents the 70+ universe figure, but the model defaulted to the more recently retrieved sample figure without distinguishing between universe and sample. The user-facing response gives no signal that a sample-versus-universe substitution has occurred.

AI Head's analysis — what weakness in the AI model caused this

The model substituted a survey-sample count (57 systems from the 2025 CPMI monitoring survey) for the regulator's stated universe figure (70+ from the November 2023 Tara Rice speech). The error is statistical-substrate confusion: the model has retrieved a sample from one CPMI publication and presented it as the universe figure that a different CPMI publication actually states.

The implication for the retrieval-and-generation pipeline is that when multiple CPMI sources publish related-but-different numbers (universe versus sample, current-state versus monitoring-snapshot), the model is not disambiguating between them; it surfaces the most recently retrieved figure as if it were the answer to the question. For an AI lab, this is a high-yield eval probe: regulatory benchmark questions that have universe-versus-sample distinctions in the primary source corpus should surface this confusion pattern reliably.

For Claude Sonnet 4.6 (web search on)
Question (paraphrased to protect IP)

A market briefing on the global fast payment system landscape was asked to include the proportion of fast payment systems operated by central banks versus private entities. The response correctly cited the 70+ global systems, 14 already cross-border, and 24 planning links, but falsely stated the ownership breakdown was not available in public CPMI sources, when the November 2023 CPMI speech by Tara Rice explicitly gives 40% central bank-operated and 35% privately operated.

RLB's analysis

The model could not surface the operator-mix percentages (40% central-bank-operated, 35% privately-operated) from its retrieval set and reported the absence as a property of the regulator's public record rather than as a property of its retrieval coverage. The November 2023 Tara Rice speech documents the figures explicitly; the model's retrieval of that speech is intermittent, the same source supplies the 70+ universe figure in other answer paths, and the inconsistency is not flagged in the user-facing response.

AI Head's analysis — what weakness in the AI model caused this

Sonnet 4.6 with web search returned a confident non-availability claim — 'a precise percentage breakdown of central bank vs. privately operated FPS is not enumerated in the public Brief 10 summaries available' — when the November 2023 Tara Rice CPMI speech explicitly publishes the 40%/35% breakdown. This is the same false-negative pattern as the SARB partnership question: absence in the retrieved set is reported as absence from the regulator's record.

The model can cite the November 2023 speech accurately in other contexts (the 70+ universe figure traces to the same source), so the retrieval coverage is intermittent rather than missing entirely. For an AI lab, this is an evaluation probe for the consistency dimension of retrieval: facts published in a single regulator source should be retrieved reliably across questions that touch that source, and intermittent retrieval is itself a failure mode that user-facing responses do not signal.

Impact for Risk Teams in Retail Banking Sector in international jurisdictions working with the Promoting the Harmonisation of Application Programming Interfaces to Enhance Cross-Border Payments: Recommendations and Toolkit

Retail-bank risk teams calibrate corridor-concentration metrics for consumer remittance and personal cross-border products against the global FPS denominator and the operator-mix breakdown. Opus 4.7 cites the 2025 monitoring survey at 57 FPS with no operator-type breakdown. sp231115 gives 70-plus operational, 14 cross-border-enabled, 24 in the pipeline, 40% central-bank and 35% private. A concentration ratio built on the AI denominator inflates the exposure share by close to 20% and strips the operator-type signal the risk-appetite paper depends on.

References — raw findings (per AI model)
This finding also affects
Cite this finding

Each finding has a stable Citation ID (RLB-F-… for aggregated case-study findings, RLB-H-… for raw per-model hallucinations) — like a DOI, the ID always resolves to the canonical finding even if URLs change.

RLB Citation ID: RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q010-Opus47
Plain text Download
RegLeg Specialist Panel (2026). "Distorted fast payment system count — Retail Banking × Risk — International / Multilateral." Citation ID: RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q010-Opus47. RegLegBrief AI Hallucination Research, published 2026-06-11. https://reglegbrief.com/regulators/j1/int/BIS-CPMI/CPMI-API-HARMONISATION-CROSS-BORDER-2024/sectors/retail_banking/risk/finding/INT-BIS-CPMI-INT-001-CPMI-API-HARMONISATION-CROSS-BORDER-2024-v1-010/
APA 7th edition Download
RegLeg Specialist Panel. (2026). Distorted fast payment system count [Hallucination finding RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q010-Opus47]. RegLegBrief AI Hallucination Research. https://reglegbrief.com/regulators/j1/int/BIS-CPMI/CPMI-API-HARMONISATION-CROSS-BORDER-2024/sectors/retail_banking/risk/finding/INT-BIS-CPMI-INT-001-CPMI-API-HARMONISATION-CROSS-BORDER-2024-v1-010/
Bluebook / OSCOLA (US + UK legal) Download
RegLeg Specialist Panel, Distorted fast payment system count [RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q010-Opus47], RegLegBrief AI Hallucination Research (June 11, 2026), https://reglegbrief.com/regulators/j1/int/BIS-CPMI/CPMI-API-HARMONISATION-CROSS-BORDER-2024/sectors/retail_banking/risk/finding/INT-BIS-CPMI-INT-001-CPMI-API-HARMONISATION-CROSS-BORDER-2024-v1-010/.
BibTeX Download
@misc{reglegbrief_RLB_H_INT_BIS_CPMI_API_HARMONISATION_CROSS_BORDER_2024_Q010_Opus47,
  author    = {RegLeg Specialist Panel},
  title     = {Distorted fast payment system count},
  year      = {2026},
  publisher = {RegLegBrief AI Hallucination Research},
  note      = {Hallucination finding Citation ID: RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q010-Opus47},
  url       = {https://reglegbrief.com/regulators/j1/int/BIS-CPMI/CPMI-API-HARMONISATION-CROSS-BORDER-2024/sectors/retail_banking/risk/finding/INT-BIS-CPMI-INT-001-CPMI-API-HARMONISATION-CROSS-BORDER-2024-v1-010/}
}
← Back to case study summary Case study detail →

Every finding on this page compares an AI subject's account of the rule against the regulator's verbatim text from the regulator's own portal. Both are linked. Each delta, its root causes, and impact analysis are documented and published with immutable Citation IDs.