Operations at a payment institution holds two CPMI-derived inputs in active use: the ISO 20022 address-format and message-validation regime (against d230), and the FPS-connectivity roadmap that sets the next 18 months of integration work (against the CPMI cross-border-payments brief series). Two AI failures on this regulation hit both inputs. Sonnet 4.6 manufactured a November 2026 structured-address cutover the d230 source does not state, and Opus 4.7 returned a 57-FPS 2025-survey count that erases the 70-plus operational figure and the operator-mix breakdown from the Tara Rice November 2023 speech.
Either error, lifted into an ops change-management ticket or the connectivity roadmap, displaces real CBPR+ and integration capacity against fabricated anchors.
What the AI got wrong, and why it matters here
Both errors land where ops normally trusts numerical specificity. Both are confidently delivered without any flag that the primary regulator text was not actually retrieved.
Finding 1: Fabricated November 2026 ISO 20022 cutover
Sonnet 4.6 committed to a November 2026 structured-address-only cutover for ISO 20022 cross-border payment messages, framed as a d230 commitment. The d230 source describes only standardisation and regulatory developments since 2023 and a separate technical annex; the November 2026 cutover is not in the document. Quoted into an address-validation pipeline change ticket, the line schedules ops capacity against a regulatory deadline that does not exist.
Citation: RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q009-Sonnet46.
Finding 2: FPS count compressed and operator mix dropped
Opus 4.7 cited the 2025 monitoring survey at 57 (56 in one graph) operational fast payment systems with no operator-type breakdown. The Tara Rice November 2023 speech (sp231115) gives 70-plus operational, 14 cross-border-enabled, 24 in the five-year planning pipeline, 40% central-bank-operated and 35% privately operated. An ops connectivity roadmap built on the AI answer is sized too small and drops the forward-pipeline signal.
Citation: RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q010-Opus47.
When this hits the ops calendar
Operations pulls CPMI material on three artefacts: the message-validation change ticket and pipeline, the correspondent-conformance schedule, and the FPS-connectivity roadmap that ops costs and product owns.
| Standing item | Where the AI risk surfaces | Failure mode |
|---|---|---|
| Address-validation pipeline change ticket | ISO 20022 cutover date commitment | Finding 1 |
| Correspondent-conformance schedule | ISO 20022 cutover commitments and FPS connectivity sizing | Findings 1 and 2 |
| FPS-connectivity roadmap | FPS count and forward-pipeline signal | Finding 2 |
Aggregate impact on the team
The fabricated cutover books address-validation capacity against a non-existent deadline; the compressed FPS count strips the forward-pipeline signal that staffing decisions depend on.
| Risk Impact | Count | Affected findings |
|---|---|---|
| 0 |
What this team should do
Tag the November 2026 cutover assertion and the 57-FPS figure as known-failure outputs. Any AI draft that contains either must be returned through a primary-source verification step against d230 and sp231115 before it lands in a change ticket or the connectivity roadmap.
Detection patterns to add to AI-review
- ISO 20022 cutover dates against d230 must be verified against the d230 text and technical annex.
- FPS counts must trace to sp231115 or to a numbered CPMI cross-border monitoring brief.
How RLB can help
RLB tracks the failure pattern on d230 and the CPMI cross-border-payments brief series and refreshes the catalogue against live AI subjects on rotation. PI ops can wire the catalogue into the change-ticket review step so these two failure shapes never reach the connectivity roadmap or the conformance schedule.