German Court Holds Google Liable for AI Overviews Errors

The Munich I Regional Court has ruled that Google bears direct liability for false content generated by the AI Overviews feature in its search engine. It is the first ruling in which a German court treated an AI system's response as the company's own statement rather than a neutral presentation of search results.

The case concerned two Munich publishers whom the AI Overviews feature linked to fraud, subscription traps, and suspicious business practices, even though this information did not appear in the sources the AI referenced. The system combined data about different, unrelated entities and presented it as facts about specific companies.

The Core of the Ruling

Central to the court's reasoning was the distinction between a classic search engine and a generative summary. A traditional search engine directs users to material published by other parties and benefits from limited intermediary liability. AI Overviews works differently: it independently compiles, synthesizes, and evaluates information from multiple sources, producing a new, original statement.

The AI overview is Google's own content, classified as a direct infringer, because the system generates independent, new, and substantive statements - from the reasoning of the Munich I Regional Court ruling

The court also stressed that the AI made claims that did not even appear in the search results underlying the summary, which the court explicitly described as the defendant's, that is Google's, own statements. The company therefore cannot hide behind the argument that it merely reflects content available online.

Google's Response

Google has said it will appeal and emphasized that the ruling concerns isolated shortcomings rather than the mechanism of the feature as such. The company argues that the vast majority of AI Overviews responses are accurate, and that errors occur only occasionally, when the system omits context or misinterprets a page's content.

We invest heavily in the quality of AI Overviews to ensure that the vast majority of responses provide accurate information, and they are designed to reflect information available on the web - Google spokesperson

Why This Precedent Matters

Until now, operators of large language models have argued that AI hallucinations are an inherent feature of the technology, making it difficult to hold anyone legally responsible. The Munich ruling breaks with that line of defense, finding that since only Google controls the algorithm and the data on which AI Overviews operates, the company is also liable for the outcome of its operation, regardless of the fact that the content is generated automatically.

The ruling currently takes the form of a preliminary injunction and is not yet final, but media law attorneys are already citing it as a reference point for further disputes in Europe. If Google's appeal does not overturn this line of jurisprudence, publishers and other companies harmed by faulty AI summaries will gain a template for lawsuits that could realistically end in a court injunction and costs borne by the model provider.

Implications for Poland and the European Union

The case was decided under German law, but its consequences extend beyond Germany, since AI Overviews operates in the same form in other European Union countries, including Poland. Polish publishers and companies that have spent months complaining about faulty summaries generated by AI search engines now have a legal argument that such content can be treated as Google's own statements rather than a neutral search result.

The case coincides with the entry into force of further obligations under the EU AI Act, which from August impose new requirements on providers of the most powerful AI models. The Munich ruling shows that alongside EU-wide regulation, national court precedents are also mounting, which could force Google to exercise greater caution when publishing AI-generated summaries.

Google now has time to file its appeal, and proceedings at the higher court could take many months. Until that appeal is resolved, the preliminary injunction covering the specific false claims made about the two Munich publishers remains in force, and the company risks a penalty of up to 250,000 euros for each violation.