German court finds Google liable for AI Overviews errors

Google liable – A Munich Regional Court ruled that Google can be held responsible for false statements generated by its AI Overviews feature, requiring the company to remove large portions of the allegedly defamatory summaries and pay 80% of legal costs. The decision draws a

The first warning signs showed up in search results, not in a chatbot window. In the Munich Regional Court’s ruling. what mattered was how Google’s AI Overviews feature presented information—linking to publishers and describing business conduct in ways the court found to be unsupported by the sources those links pointed to.

The case began after two publishers discovered that. in certain searches. Google’s AI-generated summaries connected them to questionable business practices. scams. and subscription-related frauds. They said the claims had no basis. Earlier this year, the companies sent Google a cease-and-desist letter. Google denied liability. pointing to the way the system works: its automatic summaries warn users that the information may contain errors and should be independently verified.

But the court’s reasoning went much further than a “buyer beware” disclaimer.

In its analysis. the court said Google’s AI combined information tied to other companies that had been flagged for possible illicit practices with data from the publishers involved in the case. The result was an association the court determined did not appear in any of the sources that the search engine’s links referenced.

The judges also drew a distinction between what traditional search engines do and what AI Overviews does. In the court’s view, a conventional search results page typically lists third-party links alongside statements made by others. Google’s tool, by contrast, produced what the court characterized as “independent, new, and substantial statements.”.

That difference mattered for responsibility. The court concluded that correcting misinformation is not something third parties should be forced to carry when the technology that generates the statements is controlled by the platform. Google. the court said. is the only entity able to modify the underlying system that produces AI-generated summaries. and therefore “must be held accountable.”.

The ruling also rejected a key part of Google’s defense. The court found that the challenged AI Overviews contained statements that “do not appear at all in the search results,” undermining the idea that the output was merely reflecting what was already there.

The judges further held that Google’s user-facing warning about potential hallucinations does not remove the company’s responsibility as a content distributor. In their view. accepting that logic would leave victims of false statements virtually defenseless—because the original sources never made the claims that the AI summaries attributed to them.

The court also addressed free speech arguments. It ruled that results generated by an AI system cannot be protected under free speech principles in this context. because the output is the product of an algorithm that is designed. trained. and managed by a company rather than an individual’s expression.

As a precautionary measure, the ruling required Google to remove a large portion of the statements considered defamatory in the case, and to cover 80 percent of the legal costs tied to the proceedings.

If the decision is upheld. it could shift how courts treat the role of search engines that move beyond link lists and into generative summarization. Until now. many legal systems have treated search engines as intermediaries that help users find content created by others—even when that content is false. inaccurate. misleading. or defamatory. The Munich court said that protection does not carry over the same way once generative AI enters the process and produces claims that were never explicitly made by the underlying sources.

Google’s response, at least for now, was to keep the door open to further fight. A company spokesperson—quoted by Ars Technica—said the decision could be appealed. The statement said Google “invest[s] deeply in the quality of AI Overviews to ensure that the overwhelming majority of responses provide accurate information. ” and that the summaries are designed to “reflect the information that exists on the web.” The spokesperson also said. “We’re carefully reviewing this decision. which is not yet final.”.

Google AI Overviews Munich Regional Court Germany AI hallucinations search engine liability cybersecurity publishers generative AI