In its decision dated 28 May 2026[1], the Munich Regional Court granted an injunction which had been sought by two publishers after a search engine's summary displayed several defamatory statements about their businesses.
Background
Two publishers sought injunctions against the operator of a search engine, which, in addition to displaying conventional search results, provides a feature called "AI Overview”, which summarises information from various sources and presents it to users alongside links.
The issue in respect of the two claimant publishers was that, when users entered search queries combining the claimants’ names with terms such as “fraud scheme", the search engines displayed AI Overview generated texts that associated the claimants with alleged fraudulent and deceptive business practices. The summaries suggested that the claimant publishers engaged in dubious practices such as frauds, subscription traps and improper invoicing. The overview began with a direct and definitive answer to the search query (“Yes, …”), followed by a section that listed the alleged characteristics of misconduct. A further section provided practical advice to users ("What you can do").
The claimants argued that the statements made in the AI Overview were false and damaging to their businesses.
Before initiating proceedings, the claimants requested that the defendant refrain from displaying the contested AI-generated overview text and demanded that it issue a retrospective declaration to that effect. It further submitted a complaint through the defendant's online reporting system. While the sought declaration has not been given, the AI outputs later changed and no longer contained the disputed wording.
The defendant's arguments
The defendant argued that its AI Overview merely displayed data and information from third parties in an automated manner in response to user queries. Accordingly, it argued that it was not itself a controller which could be held responsible for the processing of the data and did not adopt the third-party information as its own in the AI Overview.
Further, the defendant submitted that informed users recognised from the links to third‑party websites that the summary in the AI Overview was based on third-party content. Users could also verify via the links whether the content of the third-party websites actually corresponded to the summary given in the AI Overview – in fact, users were generally aware that this was necessary because AI-generated information "should not be trusted blindly".
In relation to the risk of repetition (something which is needed under German law for the specific injunction that was sought here) the defendant argued that there was none, as, following a revision by means of machine learning, the contested text was no longer displayed in the AI Overview when entering the above-mentioned search terms.
The claimants' arguments
The claimant publishers disagreed and maintained that the AI Overview constituted an independent statement which was attributable to the search engine operator itself, rather than a mere aggregation of third-party content. Given that the defendant had not given the sought declaration to refrain from displaying the content in dispute, the claimants argued that there remained a risk of the disputed and allegedly defamatory statements being repeated.
The Court's findings
The Munich Court sided with the claimants and held the following:
- The AI Overview in question did not merely display search results as links or with short previews (snippets); rather, the results of the search query were summarised and presented in the engine's own words and in a distinct structure. This can be seen from the introductory affirmation of the search query ("Yes,.."), which linguistically goes beyond the mere presentation of links, and continues in an independent structuring of the answer consisting of an introductory summary, followed by a compilation and a recommendation for action.
- All of this demonstrated an independent processing of search results by the AI
- Since the search engine operator itself had introduced the AI and offered it to users, the results must be attributable to it as only the search engine operator can influence the AI results and the algorithms with which the AI operates
- The AI Overview was self‑contained and readily comprehensible on its own; it conveyed a complete statement with an independently intelligible content and provided no indication of alternative interpretations or of any unreliability of its content. As a result, users had no reason to verify the answer displayed in response to their search query by consulting additional sources.
Previous German decisions in the context of search engines
In support of its position, the defendant also relied on earlier decisions of the German Federal Court of Justice ("BGH"), which the Munich court took into consideration.
In a decision from 2018,[2] the BGH held that a search engine provider could not reasonably be expected to verify, before making content discoverable, whether the content found was lawfully published, because such a proactive duty of review would contradict the purpose and functioning of a search engine. Imposing a general monitoring obligation would be difficult to implement in practice and would seriously call into question the existence of search engines as a business model.
The Munich court disagreed that the considerations made in this decision should be applied to the defendant in the present case, arguing that the AI Overview in question must be treated differently: it did not merely display search results from websites with which the search engine provider had no contact and whose content it cannot easily verify, but instead, the AI Overview produced independent, new, substantive statements based on an evaluation and combination of content from various third-party websites.
The content of these newly generated statements was capable of being reviewed by the search engine operator (and was, in fact, reviewed here). Importantly, in the present case, the third-party websites had in fact never made the statements in question about the claimant publishers, but the statements were first generated, published and disseminated by the search engine operator through its AI Overview. Consequently, any affected individuals would be unable to claim against the third parties whose websites were incorporated into the search query, because those third parties did not make the statements, leading to a liability gap.
Moreover, unlike the display of search results in search engines, an AI Overview was "by no means essential for the use of the internet"[3].
In 2013, the BGH handed down a judgment[4] concerning the issue around the “autocomplete function” when searching for terms in search engines. The BGH held that a provider was, in principle, responsible for search suggestions; however, liability was generally limited to cases in which the search engine operator becomes aware of a potential infringement caused by the autocomplete function.
The Munich court stated that the AI Overview went significantly beyond this autocomplete function and denied that the BGH's considerations should be applied. The issue in question was not the search query, the autocomplete suggestions, or the individual search results displayed via links, but rather an independently generated statement that formulated and offered its own answer based on the weighting and evaluation of search results.
Implications
A central issue of this decision will be how search engine operators will seek to control the outputs generated by their AI systems, and whether such control is realistically achievable. Given the complexity of large-scale models and the ever-changing nature of online information, ensuring consistently accurate results will be demanding. It will require ongoing financial investment in development, monitoring, and human oversight.
We assume that search engine operators may seek to reduce their liability through the use of caveats (which may be challenging due to differing applicable laws in different territories) and potentially through asking users to consent to the possibility of misleading results being generated. Arguably, even with these efforts, complete control is unlikely to be attainable. Companies may therefore face a strategic choice between heavily investing in minimising errors and accepting the risk that there will be claims arising from inaccurate or misleading outputs. Search engine operators will also have to weigh the benefits of enhancing user experience and maintaining engagement against potential financial liability.
The defendant has indicated it would appeal the decision.
[1] LG München I, Endurteil v. 28.05.2026 – 26 O 869/26
[2] BGH Judgment of 27 February 2018 – case no. VI ZR 489/16
[3] Para 40.
[4] Judgment of 14 May 2013 – case no. VI ZR 269/12
