A German court has established Google AI Overviews liability for defamatory content, classifying the feature as Google’s own speech rather than a neutral aggregation of third-party sources. The Regional Court of Munich issued the temporary injunction on 28 May 2026, in proceedings brought by two Munich-based publishers whose names had been falsely associated with subscription traps and fraudulent business practices by the AI-generated summaries (case no. 26 O 869/26).
The court found that Google’s AI Overview had confused the plaintiffs with genuinely disreputable companies and had constructed links that appeared in none of the underlying search results. One output opened with the confident assertion, ‘Yes, [company] is known for dubious business practices,’ then organised its response under self-generated headings covering red flags, alleged scam indicators, and user warnings. None of that structure or those connections appeared in the linked sources.
Google had argued that users could check the linked sources themselves to verify accuracy and that people generally understood ‘that information generated with AI should not be blindly trusted.’ The court rejected both points. The possibility of disproving a statement through further research does not, in the court’s words, ‘regularly exempt from liability for this statement.’
Google AI Overviews Liability: The Direct Disturber Classification
The legal pivot in the Munich ruling is the court’s classification of Google as an unmittelbarer Störer, or direct disturber, rather than the indirect intermediary status that has historically insulated search engines from proactive content obligations. heise online reports that this classification lowers the bar for publishers and businesses to seek preliminary injunctions against reputation-damaging AI content.
German Federal Court of Justice (Bundesgerichtshof) precedent, which the Munich court acknowledged, holds that search engines are not required to proactively check content because doing so would threaten the viability of the search engine model. ComputerBase notes the Munich court distinguished AI Overviews from that line of authority on the basis that the AI does not merely retrieve and list sources but rewrites and judges them, producing content ‘in its own words and according to its own structure.’
The court also found that only Google, not the publishers, has the technical capacity to correct the underlying algorithm and the outputs it generates. That asymmetry factored directly into the rejection of Google’s intermediary shield argument.
The false outputs were described in the ruling as ‘primarily an expression of the defendant’s commercial activity,’ and the AI’s assertions were found capable of influencing public opinion. Trending Topics reports that the court found Google’s status as an indirect interferer, a classification reinforced at EU level in part through the Digital Services Act, did not protect it in these proceedings.
Scope, Penalties, and Consequences for the Market
The injunction carries significant teeth. Violations attract a penalty of up to 250,000 euros or detention. Unusually for a domestic court order, the injunction is not limited to Germany: ComputerBase, citing The Decoder’s coverage of the ruling, reports that it applies internationally.
heise online’s English-language report notes the ruling has direct consequences for the operation of AI Overviews in the German-speaking market. For the wider answer-engine industry, the implications extend beyond Germany: Let’s Data Science reports this is believed to be the first ruling of its kind to treat AI-generated search overviews as a platform’s own speech and to apply direct liability on that basis.
The publishers had sent Google a cease-and-desist letter before proceedings began. Google’s failure to respond appropriately was a factor the court considered when issuing interim relief.
How the Analysis Compares with US Law
The Munich court’s reasoning has parallels in US jurisprudence. Under section 230 of Title 47 of the United States Code, immunity extends to intermediaries for content created by third parties, but not to platforms that are themselves responsible, in whole or in part, for the creation or development of the material in question. An AI system that assembles and outputs defamatory text, word by word, plausibly falls outside that protection on the basis that the platform has materially contributed to the alleged unlawfulness of the content.
The Munich analysis, grounded in the AI’s active rewriting and editorialising rather than passive retrieval, maps closely onto that distinction. Where the German and US frameworks are likely to diverge is on interim relief: obtaining a temporary injunction equivalent to the Munich order would, in most US jurisdictions, face a considerably higher threshold.
Subject to any onward appeal by Google, the ruling stands as an enforceable injunction with international reach. The question now is whether other jurisdictions, and the EU’s own Digital Services Act enforcement machinery, will treat the Munich reasoning as a template for regulating AI answer engines more broadly.
