A German court has ruled that Google is directly liable for false claims its AI Overviews make, treating the AI-written summaries as Google's own speech rather than ordinary search results. It is one of the first rulings to test who is responsible when a generative-AI system gets it wrong, and the answer it gives is blunt: the company that built it.
The Regional Court of Munich issued a temporary injunction barring Google from repeating false statements about two Munich publishers, whose names its AI Overviews had wrongly tied to scams, subscription traps, and “dubious business practices”.
According to the court, the AI had invented connections that appeared in none of the linked sources, mixing the publishers up with genuinely shady firms. The publishers sent a cease-and-desist letter, and Google did not respond adequately.
Not a search engine, a publisher
The crux is a legal reclassification. German search engines have long had limited liability because they merely point to third-party pages. AI Overviews, the court found, do something different: they generate “independent, new, and substantive statements” in Google's own words, so Google “alone has influence” over them and owns what they produce.
The court called the false claims “the defendant's own statements”.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!
It also rejected Google's central defence, that users can check the linked sources themselves and know not to trust AI blindly. The chance to disprove a statement through further research does not exempt whoever published it, the court said, drawing a parallel to press law, where a misleading teaser is actionable even if no one reads the full article.
Studies have found barely 1 per cent of users click a source from an AI Overview. Google could not fall back on Digital Services Act host-provider protections either.
The caveats matter. This is a preliminary injunction from a regional court, not a final judgment or binding precedent, Germany is a civil-law system, and Google can appeal. A separate German case recently dismissed a surgeon's similar claim while still affirming the principle that Google can be liable.
Google, which the court ordered to cover 80 per cent of the costs, has not commented.
The scale is why it matters beyond two publishers. An analysis for the New York Times found Google's AI Overviews, running on Gemini 3, are accurate about 91 per cent of the time, but more than half of even the correct answers were not supported by the sources cited. At Google's volume, the wrong ones add up to millions of false answers.
The same logic, if it survives appeal, would land on every AI answer engine, from ChatGPT to Perplexity, and the court said its reasoning could have international reach. It also lands amid intensifying European pressure on Google, which already faces a major EU fine and orders to open Android to AI rivals under the bloc's new AI rules. For an industry that has leaned on “AI can make mistakes” disclaimers, that is the part that should sting.