The preliminary ruling finds Google's AI Overviews produce 'new' claims that do not appear in search results, so the company's standard accuracy disclaimer does not immunize it from the false statement claims at issue.
A Munich regional court has issued a preliminary ruling holding Google liable for false statements produced by its AI Overviews feature, the AI-generated summary boxes that appear above traditional search results. The decision marks the first time a court has applied publisher-style accountability to a generative search product, and its reasoning offers a working test for who is on the hook when AI search products defame, hallucinate, or synthesize new claims.
Two publishers brought the case after Google's AI Overviews linked their names to scams, subscription fraud, and questionable business practices with no basis in fact. The companies had sent Google a cease-and-desist letter earlier in 2026 before filing the lawsuit, according to WIRED, which cited the case's originating report from Decoder. The court concluded that the false associations existed nowhere in the underlying search results. They were produced by the AI itself, by combining the publishers' data with information from companies previously flagged for possible illicit practices.
That distinction is the heart of the ruling. Search engines have long defended against defamation claims by arguing they are neutral conduits: they link to third-party content but do not produce the statements themselves. The Munich court found that AI Overviews broke that pattern. By synthesizing associations between the plaintiffs and the flagged companies, the AI produced what the ruling, as described by WIRED, characterized as "independent, new, and substantial statements" that did not exist in the linked results, WIRED reported.
Google argued that the AI's built-in disclaimer should shield it. The feature already tells users that information may contain errors and should be independently verified. The court rejected the defense on a narrow but consequential ground: the disputed statements do not appear in the search results at all, so a disclaimer about result accuracy cannot reach them. The court also pointed out that Google, as the only entity that can modify the technology underpinning the summaries, is the party that must be held accountable. Correcting the misinformation is not the plaintiffs' responsibility, WIRED reported.
The ruling orders Google to prevent further dissemination of the erroneous claims through its search engine. It is preliminary, not a final judgment on the merits, and it originates in a German regional court rather than a U.S. court or the EU's top tribunals. Google has signaled it will contest the decision. The direct precedential weight in the United States is limited, and the case will likely turn on questions of EU consumer protection, platform liability, and choice of law before any global rule emerges.
What the decision does offer is a vocabulary. Sources who say an AI product has defamed or misrepresented them now have a remedy aimed at the vendor of the AI, not just the websites the AI scraped. Courts gain a doctrinal test for distinguishing a connector of existing content from a generator of new assertions. AI search vendors have explicit notice that a generic accuracy disclaimer is not a liability shield when the disputed statements were created by the AI itself and do not exist in the linked results. The next phase of the case will test whether that reasoning survives appeal, whether other courts adopt it, and whether the same logic reaches other generative search products that synthesize rather than summarize.