This is the consolidated view of findings. Click the Citation IDs or 'see details →' on any item for the full details for each finding.
AI assistants we tested invented a detailed four-area toolkit structure — complete with named assessment dimensions — that no accessible public source supports. A Technology & Data team that scopes an API harmonisation readiness exercise using this AI output will produce a gap analysis mapped to fabricated criteria, which then circulates through internal sign-off and potentially into partner due-diligence documentation before anyone verifies the primary BIS source.
For a SaaS firm positioning its payment API stack against CPMI expectations, a readiness report built on non-existent criteria creates remediation cost and, in a partner technical review, a credibility exposure that is difficult to walk back.
AI assistants we tested cited an incorrect publication date for the February 2026 ISO 20022 data requirements update, sourcing an April 2026 third-party aggregator date rather than the primary BIS record, and simultaneously fabricated specific data entity breakdowns for the technical annex. A technical briefing or engineering roadmap item that carries the wrong version date misanchors the implementation timeline — teams may treat the wrong month as the operative date for when updated requirements apply, and may scope to annex-level entity obligations that the document does not actually contain.
For a Software & SaaS firm whose product touches ISO 20022 message flows, version attribution errors in internal technical documentation propagate into engineering specifications before the discrepancy surfaces.