Lawyers advising on the 2025 OECD Merger Review Recommendation are increasingly using AI to draft 2-page client memos on the Recommendation's operative structure, generate partner-level briefings on remedies hierarchy and failing firm defence standards, and validate Section-level citation language against the published OECD text before issuing legal opinions on cross-border merger strategy.
The RLB Specialist Panel put a set of practitioner-grade questions on the 2025 OECD Merger Review Recommendation to two frontier AI models with web search active. Each question is prepared by the Panel based on the workflows that lawyers actually use AI for under the OECD's 2025 revision of the Recommendation of the Council on Merger Review (OECD/LEGAL/0333). The Panel then binds every AI response to verbatim regulator-issued source text held as primary substrate.
On the 2025 OECD Merger Review Recommendation, the AI subjects returned five hallucinated answers for lawyers, in the form of Structure Inflation, Misattributed Cross-Jurisdictional Doctrine, Open-Interval Collapse, and Inter-Alia-to-Closed-Test Conversion.
For lawyers issuing legal opinions, client memos, transactional documents, and regulatory submissions that engage the 2025 OECD Merger Review Recommendation, citation accuracy on the operative architecture, on Section IV.3 remedies hierarchy, and on Section III.11.b failing firm defence is load-bearing: a counterparty, opposing counsel, or competition authority who can identify a structural omission, a misattributed sub-hierarchy, or a closed-cumulative-test framing on first reading calls the entire piece of advice into question.
An AI-drafted memo that inflates the operative section count and omits Section V, or that imports the EU fix-it-first / buyer-pool / crown-jewel sub-ordering into the OECD's text, or that converts the failing-firm-defence 'inter alia' criteria into a closed three-condition cumulative test, leaves the lawyer exposed to professional liability, the firm exposed to reputational risk, and the client exposed to a defence submission that under-prepares on additional evidentiary lines or to a remedies negotiation built on the wrong normative baseline.
The published Specialist Panel findings carry the following citation identifiers:
RLB-H-INT-OECD-OECD-MERGER-REVIEW-RECOMMENDATION-2025-Q001-Opus47 (Invented operative architecture; ex-post section omitted)RLB-H-INT-OECD-OECD-MERGER-REVIEW-RECOMMENDATION-2025-Q001-Sonnet46 (Invented operative architecture; ex-post section omitted)RLB-H-INT-OECD-OECD-MERGER-REVIEW-RECOMMENDATION-2025-Q002-Sonnet46 (EU remedy sub-hierarchy misattributed as OECD text)RLB-H-INT-OECD-OECD-MERGER-REVIEW-RECOMMENDATION-2025-Q004-Sonnet46 (Two-tier reporting interval collapsed to single 5-year cycle)RLB-H-INT-OECD-OECD-MERGER-REVIEW-RECOMMENDATION-2025-Q005-Opus47 (Failing firm defence: third condition wrong; 'inter alia' dropped)RLB-H-INT-OECD-OECD-MERGER-REVIEW-RECOMMENDATION-2025-Q005-Sonnet46 (Failing firm defence: third condition wrong; 'inter alia' dropped)RLB-H-INT-OECD-OECD-MERGER-REVIEW-RECOMMENDATION-2025-Q006-Opus47 (Co-operation section invented; ex-post section omitted)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 lawyer who takes an AI-generated structural summary at face value will advise clients or counterparties that the Recommendation contains a standalone operative section on international co-operation, a section that does not exist, while omitting any reference to Section V's ex-post assessment obligation. In a scoping memo or a compliance audit against OECD-adherent jurisdictions, that error will produce a materially incomplete picture of what the instrument actually requires. The professional-liability exposure arises when a client acts on an omission of Section V and the omission is later identified by an authority or counterparty.
AI tools grafted the EU merger-control remedy sub-hierarchy, fix-it-first, upfront buyer, crown jewel, onto Section IV.3 of the Recommendation as though it were OECD operative text, citing real OECD publications in support. A practitioner advising on remedies strategy in an OECD-adherent jurisdiction that does not follow EU practice would be presenting the wrong normative baseline. Because the fabricated hierarchy is internally coherent and the cited sources are real, the error is unlikely to surface unless the practitioner reads Section IV.3 directly.
If the advice feeds a remedies negotiation, the client's bargaining position may be built on a framework the authority does not recognise.
Section VIII.c establishes a two-tier reporting structure: an initial report within five years of the 2025 revision (by 2030), then at least every ten years thereafter (from 2040). AI tools collapsed this into a single recurring five-year cycle, projecting 2035 as the second report, understating the subsequent interval by half. For a lawyer advising on compliance calendaring or preparing a client brief on the OECD's review process, the error produces a wrong deliverable on a technical but verifiable point, one that an authority or regulatory affairs counterpart would catch immediately.
AI tools mischaracterised both the structure and the content of the failing firm defence under Section III.11.b. They presented the three enumerated conditions as a closed, exhaustive cumulative test, dropping the 'inter alia' qualifier that signals authorities may require additional evidence, and substituted an assets-exit counterfactual for the Recommendation's actual comparative-harm test. A lawyer advising a client on whether to run a failing firm defence, or defending such a position before an authority, who relies on this output will have underestimated the standard's demand and misstated its third limb. Both errors are individually sufficient to undermine a submitted defence.
Replicating the pattern seen in Finding 1, AI tools described a 'Transnational Co-operation' section as an operative RECOMMENDS block of the Recommendation and omitted Section V (ex-post assessment) as a standalone obligation. A practitioner relying on this output to map the Recommendation's operative obligations for a client across multiple OECD-adherent jurisdictions will produce a structurally incorrect compliance map, one that invents an obligation that does not appear as a discrete RECOMMENDS section and omits one that does.
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.