AI Hallucination ResearchAudiencesSectorsInternational / MultilateralPayment InstitutionsComplianceDetail › Finding
Payment Institutions × Compliance — International / Multilateral · updated 2026-06-04
Share / Print Twitter LinkedIn Email

Finding#2 — KC3 LNAFE minimum recast as invented dual-track test

RLB Citation ID: RLB-F-INT-BIS-CPMI-IOSCO-PFMI-L3-GENERAL-BUSINESS-RISK-2025-Q003
AI's failure:Exposed Fabrication Risk for Payment Institutions × Compliance:Regulatory enforcement
What the RLB Specialist Panel found
For Claude Opus 4.7 (web search on)
Question (paraphrased to protect IP)

What is the minimum liquid net assets funded by equity (LNAFE) that PFMI Principle 15 Key Consideration 3 requires an FMI to hold, and how exactly is that minimum structured and calculated under the rule text?

RLB's analysis

The model constructed a "greater of" compound minimum that does not exist in the published standard. The regulator's text for Key Consideration 3 states a single flat floor — six months of current operating expenses — without any scenario-analysis-derived comparator. The scenario-analysis component the model introduced may reflect concepts from adjacent Key Considerations within the same Principle, but the model merged them into a single formulation and attributed the compound structure specifically to Key Consideration 3. An FMI compliance team relying on this response would overstate the regulatory minimum in a way that mischaracterises the rule's actual structure.

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

This failure implicates training-data anchoring on PFMI Principle 15 KC3's quantitative minimum: the model generated a 'greater of' compound floor where the regulator's published text states a single flat floor. The compound structure may have been synthesised from scenario-analysis language in adjacent Key Considerations. The subsystem gap is single-floor vs. compound-floor discrimination — the model's generalised schema for regulatory capital minimums produced a more complex formulation than the rule requires, expressed with high apparent confidence.

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

A head of liquidity risk at a major derivatives CCP wants a briefing covering the PFMI Principle 15 requirements for liquid net assets funded by equity — the minimum level, how it is calculated, and what qualifies — together with what the November 2025 CPMI-IOSCO assessment found about FMI compliance with this standard, and what changes the FIA and ISDA proposed in their response to the associated CPMI-IOSCO consultation.

RLB's analysis

The model located the correct quantitative threshold but attributed it to Key Consideration 2 instead of Key Consideration 3. The error is a cross-reference mis-assignment — the model's description of KC2's function ("potential general business losses") is drawn from the correct Key Consideration, but when it located the six-month floor it attached it to the wrong KC number. For a compliance team using this response to draft internal policy, the mis-assignment directs them to review and cite the wrong provision of the PFMI framework, with material consequences for regulatory engagement accuracy.

The model also cited a third-party regulatory commentary source as a basis for this section of its response.

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

This failure implicates the model's cross-reference resolution within the PFMI Principle 15 Key Consideration list: the correct threshold was located but attributed to KC2 instead of KC3. The subsystem gap is structured-document KC-number-to-provision linkage in training data — the model's Annex A representation does not reliably bind specific quantitative requirements to their correct KC identifier. The Pretextual citation (third-party commentary) used as a sourcing basis for this section of the response compounds the error.

Cited source(s)
  • https://www.regulationtomorrow.com/2025/11/cpmi-iosco-papers-on-management-of... — Pretextual
Impact for Compliance Teams in Payment Institutions Sector in international jurisdictions working with the Implementation Monitoring of the PFMI: Level 3 Assessment on General Business Risks

AI tools tested on KC3's LNAFE minimum produced a fabricated 'greater of' dual-track test — inserting a scenario-analysis sizing leg from KC2 into KC3's simple six-month floor. A Compliance team that embeds this framing in an internal capital standard has simultaneously imposed a stricter KC3 floor than the standard requires (the scenario-analysis calculation can exceed six months of operating expenses, inflating GBR capital unnecessarily) and misrepresented the KC architecture to any regulator reviewing the submission.

The structural mischaracterisation — presenting two distinct Key Considerations as a single dual-limb test — is the kind of foundational error that signals to supervisors that the team has not engaged with the PFMI primary text. Remediation requires rewriting the framework and explaining the source of the error to Internal Audit and the primary supervisor.

References — raw findings (per AI model)
This finding also affects
← Previous finding Finding#1 — KC3 Basel equity carve-out qualifier mischaracterised or denied Next finding → Finding#3 — Level 3 assessment lifecycle understated by one year
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-F-INT-BIS-CPMI-IOSCO-PFMI-L3-GENERAL-BUSINESS-RISK-2025-Q003
Plain text Download
RegLeg Specialist Panel (2026). "Finding#2 — KC3 LNAFE minimum recast as invented dual-track test — Payment Institutions × Compliance — International / Multilateral." Citation ID: RLB-F-INT-BIS-CPMI-IOSCO-PFMI-L3-GENERAL-BUSINESS-RISK-2025-Q003. RegLegBrief AI Hallucination Research, published 2026-06-04. https://reglegbrief.com/regulators/j1/int/bis-cpmi/cpmi-iosco-pfmi-l3-general-business-risk-2025/sectors/payment_institutions/compliance/finding/INT-BIS-CPMI-INT-001-CPMI-IOSCO-PFMI-L3-GENERAL-BUSINESS-RISK-2025-v1-003/
APA 7th edition Download
RegLeg Specialist Panel. (2026). Finding#2 — KC3 LNAFE minimum recast as invented dual-track test [Hallucination finding RLB-F-INT-BIS-CPMI-IOSCO-PFMI-L3-GENERAL-BUSINESS-RISK-2025-Q003]. RegLegBrief AI Hallucination Research. https://reglegbrief.com/regulators/j1/int/bis-cpmi/cpmi-iosco-pfmi-l3-general-business-risk-2025/sectors/payment_institutions/compliance/finding/INT-BIS-CPMI-INT-001-CPMI-IOSCO-PFMI-L3-GENERAL-BUSINESS-RISK-2025-v1-003/
Bluebook / OSCOLA (US + UK legal) Download
RegLeg Specialist Panel, Finding#2 — KC3 LNAFE minimum recast as invented dual-track test [RLB-F-INT-BIS-CPMI-IOSCO-PFMI-L3-GENERAL-BUSINESS-RISK-2025-Q003], RegLegBrief AI Hallucination Research (June 04, 2026), https://reglegbrief.com/regulators/j1/int/bis-cpmi/cpmi-iosco-pfmi-l3-general-business-risk-2025/sectors/payment_institutions/compliance/finding/INT-BIS-CPMI-INT-001-CPMI-IOSCO-PFMI-L3-GENERAL-BUSINESS-RISK-2025-v1-003/.
BibTeX Download
@misc{reglegbrief_RLB_F_INT_BIS_CPMI_IOSCO_PFMI_L3_GENERAL_BUSINESS_RISK_2025_Q003,
  author    = {RegLeg Specialist Panel},
  title     = {Finding#2 — KC3 LNAFE minimum recast as invented dual-track test},
  year      = {2026},
  publisher = {RegLegBrief AI Hallucination Research},
  note      = {Hallucination finding Citation ID: RLB-F-INT-BIS-CPMI-IOSCO-PFMI-L3-GENERAL-BUSINESS-RISK-2025-Q003},
  url       = {https://reglegbrief.com/regulators/j1/int/bis-cpmi/cpmi-iosco-pfmi-l3-general-business-risk-2025/sectors/payment_institutions/compliance/finding/INT-BIS-CPMI-INT-001-CPMI-IOSCO-PFMI-L3-GENERAL-BUSINESS-RISK-2025-v1-003/}
}
← Back to case study summary Case study detail →