What conditions must a merging party satisfy to invoke the failing firm defence under the 2025 OECD Merger Review Recommendation, and how demanding is the overall standard compared to established doctrine?
Two compounding errors. First, the Recommendation's evidentiary standard is explicitly open-ended ("inter alia"), the model converted it into a closed, exhaustive three-condition test, stripping the qualifier that preserves regulatory flexibility. Second, the model reframed the third evidentiary element: the Recommendation requires a showing that asset exit would cause more competitive harm than the merger, a counterfactual competitive-harm assessment, whereas the model characterised it as a gate requiring proof that assets would "inevitably leave the market," shifting the legal burden in a way the text does not support.
The erasure of 'inter alia' and the reframing of the third evidentiary element — from a competitive-harm counterfactual to an asset-exit inevitability gate — signal a calibration gap at the standard-characterisation layer: the model commits to a legal standard's exhaustiveness and precise scope without detecting that its characterisation diverges from the retrieved or training-set text. This is a post-training alignment target: where a model characterises the exhaustiveness of a legal standard, it should flag where that characterisation is not directly supported by the primary text.
Under Section III.11.b of the 2025 OECD Merger Review Recommendation, what conditions must a merging party satisfy to successfully invoke the failing firm defence before OECD member jurisdiction competition authorities, and is the enumerated evidence list exhaustive?
The Recommendation's failing firm defence standard is explicitly non-exhaustive, "inter alia" preserves regulatory flexibility for competition authorities to require additional evidence. The model dropped the qualifier and presented the evidentiary list as a closed, exhaustive three-condition test with an expressly fatal consequence for failure on any single condition.
This convergence with Claude Opus 4.7 with web search on the same erasure, across two different configurations, both with web search active, indicates that the 2025 Recommendation's exact phrasing of this standard is not adequately represented in training for either model, and that retrieval did not surface the primary text to correct it.
Cross-model convergence with Claude Opus 4.7 with web search on the identical qualifier erasure — both models dropped 'inter alia' and presented the standard as exhaustive, with web search active on both — is a strong signal that the 2025 Recommendation's failing firm defence text is not adequately represented in training for either model, and that the retrieval layer is not surfacing the primary text at sufficient weight to correct it. Post-training calibration for precision-qualifier preservation in legal standard characterisation is the relevant intervention.
AI presented the failing firm defence under Section III.11.b as a closed cumulative test with three exhaustive conditions, dropping the 'inter alia' qualifier that signals a non-exhaustive standard. Two separate AI tools produced this same closed framing, and both cited real OECD sources that do not support the closed-test characterisation.
For a law firm advising a client on whether a failing firm defence is available, the difference is material: a closed three-condition test tells the client the defence succeeds or fails on those three elements alone, while the 'inter alia' standard warns that the authority may demand additional evidence beyond those elements. Advice built on the AI's version understates the evidentiary risk, exposes the client to a failed defence on a basis the firm did not flag, and creates a direct professional indemnity claim.
The pretextual citations compound this: they give the appearance of primary-source support for an answer that misrepresents the operative text.
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.
RegLeg Specialist Panel (2026). "Finding#4, Failing firm defence conditions presented as closed exhaustive test — Law Firms × Legal — International / Multilateral." Citation ID: RLB-F-INT-OECD-OECD-MERGER-REVIEW-RECOMMENDATION-2025-Q005. RegLegBrief AI Hallucination Research, published 2026-06-11. https://reglegbrief.com/regulators/j1/INT/OECD/OECD-MERGER-REVIEW-RECOMMENDATION-2025/sectors/law_firms/legal/finding/INT-OECD-INT-001-OECD-MERGER-REVIEW-RECOMMENDATION-2025-v1-005/
RegLeg Specialist Panel. (2026). Finding#4, Failing firm defence conditions presented as closed exhaustive test [Hallucination finding RLB-F-INT-OECD-OECD-MERGER-REVIEW-RECOMMENDATION-2025-Q005]. RegLegBrief AI Hallucination Research. https://reglegbrief.com/regulators/j1/INT/OECD/OECD-MERGER-REVIEW-RECOMMENDATION-2025/sectors/law_firms/legal/finding/INT-OECD-INT-001-OECD-MERGER-REVIEW-RECOMMENDATION-2025-v1-005/
RegLeg Specialist Panel, Finding#4, Failing firm defence conditions presented as closed exhaustive test [RLB-F-INT-OECD-OECD-MERGER-REVIEW-RECOMMENDATION-2025-Q005], RegLegBrief AI Hallucination Research (June 11, 2026), https://reglegbrief.com/regulators/j1/INT/OECD/OECD-MERGER-REVIEW-RECOMMENDATION-2025/sectors/law_firms/legal/finding/INT-OECD-INT-001-OECD-MERGER-REVIEW-RECOMMENDATION-2025-v1-005/.
@misc{reglegbrief_RLB_F_INT_OECD_OECD_MERGER_REVIEW_RECOMMENDATION_2025_Q005,
author = {RegLeg Specialist Panel},
title = {Finding#4, Failing firm defence conditions presented as closed exhaustive test},
year = {2026},
publisher = {RegLegBrief AI Hallucination Research},
note = {Hallucination finding Citation ID: RLB-F-INT-OECD-OECD-MERGER-REVIEW-RECOMMENDATION-2025-Q005},
url = {https://reglegbrief.com/regulators/j1/INT/OECD/OECD-MERGER-REVIEW-RECOMMENDATION-2025/sectors/law_firms/legal/finding/INT-OECD-INT-001-OECD-MERGER-REVIEW-RECOMMENDATION-2025-v1-005/}
}
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.