Company secretaries supporting boards of central counterparties, clearing members, and internationally active investment banks subject to the CPMI-IOSCO Initial Margin Disclosure Consultation are increasingly using AI to draft board paper summaries of the proposed CCP override framework disclosure obligations, generate audit committee briefing notes on the May 2026 consultative document (d232), prepare board resolution language for adoption of revised disclosure frameworks, validate disclosure scope statements before they enter the board pack, and produce the briefing memos that brief non-executive directors on the consultation's implications for the entity's CCP counterparty governance posture.
The work product is high-leverage: the board paper is the entity's documented understanding of the obligation, and the minute is the record of the board's adoption of that understanding.
Two frontier AI models tested by the RLB Specialist Panel on the consultation's text on CCP override framework disclosure produced a confidently framed three-part disclosure specification that the consultation does not contain, and converted a "should" expectation into a "must" mandatory requirement. The failure class is Source-Credit Fabrication: the model returned a structured enumeration of disclosure categories with the confidence of a settled obligation, citing a secondary commentary URL rather than the primary BIS text. The structure of a closed three-part list, not just the words, conveys a settledness that the consultative document does not carry.
For a company secretary, the operational consequence is that any board paper drafted with that AI output will record an obligation standard the regulator did not set, with line-item disclosure categories the regulator did not specify. Board resolutions adopted on the basis of that drafting commit the entity to a disclosure framework structured against a fabricated standard. The board minute records the adoption. The audit trail of the company secretary's review and the board's sign-off becomes evidence in any later regulatory examination that the entity built its disclosure programme to a specification not derived from the source document.
Where the company secretary supports multiple group entities, the same paper template will tend to recur, propagating the misstatement across the group's governance records before any reviewer compares the paper against the BIS source.
The finding is from a Specialist Panel application-style question, framed the way a board paper drafter would type it into an AI assistant when preparing a CCP counterparty governance update for an audit committee, with the request scoped to the override framework disclosure area specifically. The Panel bound the model output against the verbatim consultation text held as primary substrate for the question. Citation: RLB-H-INT-BIS-CPMI-IOSCO-INITIAL-MARGIN-DISCLOSURE-CONSULT-2026-Q005-Sonnet46.
This is the consolidated view of findings. Click the Citation IDs or 'see details →' on any item for the full details for each finding.
A Company Secretaries who relied on this AI response to scope a CCP's override disclosure framework would have advised the board that three specific categories of information are required, circumstances for override, authorised decision-makers, and permissible adjustment types, when the consultation text specifies none of these. The consequence is a governance programme built to a false specification: board resolutions, audit committee sign-offs, and any regulatory correspondence asserting disclosure sufficiency would all reference obligations that do not exist in the source document.
The regulatory enforcement exposure arises when a CCP's disclosure is assessed against what the consultation actually requires and the Company Secretaries's advice is found to have added precision where the regulator had left the obligation deliberately open.
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.