Munich court rules Google liable for false AI Overview summaries and must delete defamatory claims

A regional court in Munich has ruled that Google is legally responsible for false information generated by its “AI Overview” feature (Übersicht mit KI) within Google Search. The decision, issued by the 26th Civil Chamber specialized in press and speech law, could set a significant precedent for the liability of AI-generated content.

The false allegations

Two Munich-based publishers sued Google after the AI Overview falsely linked them to fraud schemes, subscription traps and unscrupulous business practices. The system blended data about genuinely dubious companies with the plaintiffs and invented connections that did not exist in any of the linked sources. One of the publishers was accused of tricking users into paid subscriptions, withholding paid digital content and repeatedly demanding payments already received. The defamatory summaries appeared when users searched for the publisher's name alongside the word “Betrugsmasche” (fraud scheme). When the publisher asked Google to correct the error, the company did not respond.

“We invest heavily in the quality of AI Overviews to ensure that the vast majority of answers provide correct information.”

Court: AI summaries are Google's own content

Google had argued that the AI Overview merely reflects third-party content and that the company should not be held liable as a mere intermediary. The court rejected this position. It held that the AI-generated summary goes beyond displaying or linking to search results: it rephrases, evaluates and structures information into a self-contained statement with independently understandable content. Because Google creates entirely new assertions, it cannot rely on existing Federal Court of Justice case law that shields search engines from direct liability for listing third-party content.

The judges also dismissed Google's argument that users could verify the sources themselves or know that AI-generated information should not be trusted blindly. The chamber emphasised that the AI Overview presents itself as a closed statement without any visible warning of unreliability. The possibility of double-checking links does not release Google from liability for reputational harm.

Injunction and precedent

The court issued an interim injunction ordering Google to cease spreading the false statements and to pay 80 percent of the legal costs. The ruling is not yet legally binding, as Google can appeal. A few months earlier, a Hamburg court reached the same conclusion in an unreported, similar case.

Legal observers describe the Munich decision as potentially groundbreaking. It joins a growing debate on whether providers of generative AI systems must answer for their models' hallucinations (statements that sound plausible but are factually wrong). Across many countries, the rules governing such liability remain unsettled.

The Munich ruling makes clear that when an algorithm writes summaries, it is no longer a messenger but an author, and authors must stand behind their words.