Compliance leads in retail banking touch CPMI material when the consumer-protection lens lands on a cross-border or remittance product: which corridors are about to see a regulator-led pre-validation rail, which d224 obligations actually bind the retail bank versus the system operator the bank integrates with, and how ISO 20022 address fields change for sanctions and travel-rule screening.
Three AI-generated answers tested against d224 and the CPMI brief series were confidently wrong on those three operational signals: Opus 4.7 denied the SARB partnership documented in CPMI Brief No. 9, Opus 4.7 invented a per-recommendation stakeholder taxonomy, and Sonnet 4.6 manufactured a November 2026 ISO 20022 cutover the d230 text does not state. Each error lands inside a compliance deliverable the supervisor, the conduct regulator or the consumer-protection committee will eventually read.
What the AI got wrong, and why it matters here
All three failures share the same pattern: confident output on regulator-source detail that the AI could not actually retrieve, written in the same surface tone as a verified retrieval.
Finding 1: SARB pre-validation partnership denied
Opus 4.7 denied any public CPMI statement names SARB as the pre-validation pilot partner. CPMI Brief No. 9 (November 2025) names SARB outright. Quoted into a South-Africa-corridor remittance fairness memo or a supervisor thematic response, the denial misses a documented regulator workstream.
Citation: RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q007-Opus47.
Finding 2: Invented per-recommendation stakeholder taxonomy
Opus 4.7 returned a clean stakeholder taxonomy across d224's 10 recommendations, built from category labels rather than the recommendation text. A retail-bank compliance scoping memo written off that taxonomy misallocates obligation between the bank and the FPS scheme operator.
Citation: RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q008-Opus47.
Finding 3: Fabricated November 2026 ISO 20022 cutover
Sonnet 4.6 committed to a hard November 2026 structured-address-only cutover for ISO 20022 cross-border payment messages. The d230 source describes only generalised standardisation and regulatory developments since 2023 and a separate technical annex. A remittance product disclosure review or sanctions-screening recalibration that quotes the AI line schedules work against a non-existent deadline.
Citation: RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q009-Sonnet46.
When this hits the retail-bank compliance calendar
Retail-bank compliance pulls CPMI material on the remittance product fairness review, the consumer-protection horizon scan, the supervisor TR-7-style thematic response, and the periodic financial-promotions review.
| Standing item | Where the AI risk surfaces | Failure mode |
|---|---|---|
| Remittance product fairness review | All three | All three |
| Consumer-protection horizon scan | Pilot-partner naming | Finding 1 |
| Supervisor TR-7-style thematic response | Obligation mapping | Finding 2 |
| Periodic financial-promotions review | ISO 20022 cutover commitments | Finding 3 |
Aggregate impact on the team
The three failures spread across the supervisor-facing paper, the obligation-mapping memo and the remittance product review. The downside is supervisory and conduct-regulator exposure on factual error.
| Risk Impact | Count | Affected findings |
|---|---|---|
| 0 | ||
| 0 |
What this team should do
Treat AI output naming a CPMI pilot partner, mapping d224 recommendations to retail-bank obligations, or asserting an ISO 20022 cutover date as draft material requiring verification against the primary regulator text (CPMI brief by number, d224 recommendation text, d230 source).
Detection patterns to add to AI-review
- Pilot-partner naming must trace to a numbered CPMI brief.
- Obligation mapping on d224 must be verified against the recommendation text.
- ISO 20022 cutover dates against d230 must be verified against the d230 text.
How RLB can help
RLB tracks AI failures on d224, d230 and the CPMI brief series and refreshes the catalogue against live AI subjects on rotation. Retail-bank compliance can wire the catalogue into the AI-draft review step so these three failure shapes are caught before the language ships to the supervisor or the consumer-protection committee.