The Regional Court Munich I (Landgericht München I) issued a temporary injunction on 28 May 2026 (Ref. 26 O 869/26) ordering a search engine operator to cease publishing AI-generated overviews that falsely associated two Munich-based publishing companies with subscription traps and scams. The AI Overview feature had retrieved content from multiple third-party websites and synthesised it into summaries attributing to the publishers conduct they had not engaged in. The injunction is enforceable immediately, with fines of up to €250,000 per violation for continued publication.
The court classified the operator as a direct Täter (primary wrongdoer) rather than a Störer (indirect disturber), applying §§ 823 Abs. 1 and 1004 BGB (delictual interference and cessation claim) and §§ 3, 4 UWG (prohibition of misleading commercial practices). The court held that AI Overviews produce independent, new, and substantive statements distinct from the underlying source material, and that only the operator can verify those statements by comparing the AI output against the crawled source websites. On this basis, the operator cannot invoke the hosting safe harbour under §§ 7-10 of the Telemediengesetz or Article 6 of the Digital Services Act, which apply to hosts that do not themselves generate the unlawful content.
The decision creates liability exposure for any operator running an AI system that retrieves third-party content and synthesises it into user-facing statements. Retrieval-augmented generation (RAG) systems, generative search tools, AI-powered financial research aggregators, and AI advisory tools that summarise contractual or product terms all operate under the same attribution logic applied by the court. Financial institutions using AI to generate client-facing summaries of market data, third-party research, or credit analysis should audit their output pipelines for statements that could be classified as operator-authored rather than attributed quotations, because the operator bears liability for false AI-synthesised conclusions.
The ruling is an interim injunction (einstweilige Verfügung) and may be appealed to the Oberlandesgericht München. Full proceedings on damages have not been filed. The decision applies German law and does not bind courts in other EU Member States, though courts applying comparable direct-liability tests under national tort or unfair competition law may reach similar conclusions. The question of whether Article 6 of the Digital Services Act protects AI-synthesised outputs from hosting liability has not been addressed by the Court of Justice of the EU.
Licentium advises AI operators, regulated institutions, and technology companies on AI-generated content liability across EU and UK jurisdictions. Our practice covers AI output risk assessment, pre-deployment legal review of RAG and generative search pipelines, and defamation and unfair competition exposure analysis. Work we undertake includes AI liability governance, EU AI Act Article 13 and 22 compliance review, Digital Services Act safe-harbour analysis, and enterprise AI deployment governance.