Munich Court Holds Google Liable for False AI Overviews

A German court has ruled that Google is directly responsible for false statements produced by its AI Overviews, the short answers that appear at the top of search results. The Regional Court of Munich I issued the decision on May 28, 2026, in a case brought by two publishers, and legal commentators have since studied it closely. This article summarises the ruling using a law firm briefing from DAC Beachcroft, the legal journal MediaLaws and a case record, explains the court's reasoning, and notes what remains open. It is a summary of public information and is not legal advice, and the case is still at the interim stage.

What happened

Two publishers found that when users searched for their names together with terms such as fraud scheme, Google's AI Overview linked them to alleged fraud, deceptive practices, subscription traps and improper invoicing. According to the DAC Beachcroft briefing, the overview began with a definitive answer, followed by a list of alleged misconduct and advice under the heading of what the reader can do. MediaLaws reports that the alleged link to fraud could not be traced to any identifiable source, even though one cited article never mentioned the publishers.

Before going to court, the publishers asked the search operator to stop displaying the text and to give a declaration that it would not repeat it, and they used its reporting system. The operator did not give the declaration, although the AI outputs later changed and no longer contained the disputed wording. The court granted a preliminary injunction on May 28, 2026 in case 26 O 869/26. A digital policy database lists the parties as Publishing House 24 and Geramont publishers against Google.

The court's reasoning

The central finding was that the AI Overview is the operator's own content. The court said the overview summarised search results in the engine's own words and structure, going beyond links or short snippets. Because the AI processed the material independently, the output was attributable to the company that introduced the feature and controls its algorithms. MediaLaws says this puts Google closer to a publisher than to a neutral intermediary.

The court therefore distinguished earlier decisions of Germany's Federal Court of Justice, in 2013 on search autocomplete and in 2018 on search liability, which limited responsibility for content from third parties. A traditional search engine surfaces content without adding to it, the court said, while the AI generated new statements that no third-party site had made. That left the affected publishers with nobody else to sue, because the websites cited could not be held responsible for claims they never made.

The arguments the court rejected

Google argued that the overview only automatically displayed third-party information and that it had not adopted the content as its own. It argued that users could see from the links that the summary came from third parties and could check it, since AI output should not be trusted blindly. It argued that there was no risk of repetition because machine learning revisions had removed the text. The court rejected these points. It found that the overview was self-contained and gave no sign of unreliability, so users had no reason to check other sources, and the risk of repetition was met, because the publishers had not received the declaration they asked for.

MediaLaws adds that the court concluded Google could not rely on the hosting provider protections in the European Digital Services Act, which cover content created by others, not content a service generates itself.

The remedy and its reach

The court prohibited Google from repeating the specific false claims, with fines of up to EUR 250,000 for each violation, according to MediaLaws. The article states that the court said the ruling would apply beyond Germany under Article 36(1) of the Brussels I Regulation, which governs the recognition of judgments within the European Union. DAC Beachcroft reports that Google has said it will appeal, and MediaLaws reports that Google disagrees and describes the case as narrow. No appeal decision has been reported.

The wider regulatory picture

MediaLaws lists related action in Europe and the UK. On April 29, 2026, the Italian regulator AGCOM referred AI Overviews and AI Mode to the European Commission under the Digital Services Act. The European Publishers Council filed an antitrust complaint, the Commission has an open investigation into how publisher content feeds Google's AI features, and on June 3, 2026, the UK Competition and Markets Authority imposed a requirement letting publishers opt out of AI features without losing their standard search ranking.

What it could mean for insurers and other businesses

The ruling does not concern insurers, and the sources reviewed do not discuss them. The connection is one of principle. An insurer that puts a generative AI chatbot in front of customers, or uses AI to summarise claims, could face a similar argument that the output is its own statement, and a disclaimer that users should verify the answer may not be enough. DAC Beachcroft says operators will need continued investment in development, monitoring and human oversight, and that complete control is unlikely. It also notes that caveats or consent notices are possible but may conflict with the laws of different countries. These are legal commentators' views, not holdings of the court, and a higher court may decide differently.

What to watch next

Watch for the appeal, for any decision in the main proceedings, for the Commission's handling of the AGCOM referral and the publishers' complaint, and for whether other courts adopt the same reasoning. This article summarises public reporting and is not legal advice.