A Munich court has ruled that Google bears full legal responsibility for false claims made by AI Overviews, rejecting the company’s argument that its generative answer block merely surfaces what the web already says. The Regional Court of Munich (Landgericht München, case 26 O 869/26) banned specific defamatory statements and ordered Google to pay 80 percent of the plaintiffs’ legal costs. The decision is among the first major rulings in any jurisdiction to treat a generative search feature as authored speech rather than an indexed reference.

The facts that triggered the case are exact and important. Google’s AI Overviews falsely associated two Munich-based publishers with scam operations, subscription traps, and shady business practices. Statements included confident assertions such as “Yes, [company] is known for dubious business practices.” None of the source pages linked in the Overview contained those associations. The AI generated the reputational damage independently.

The court’s reasoning is the part SEO and GEO teams need to read carefully. The judges drew a firm line between traditional search results and AI Overviews. Traditional results point to third-party pages; the publisher of those pages carries responsibility for the content. AI Overviews do something different. The court found that Google’s system “rewrites and judges results in its own words and according to its own structure,” producing “independent, new, and substantive statements.” Because Google built the system and offered it to users, the court held, Google is the author of those statements. The liability shield that normally protects search operators who aggregate third-party content does not apply when the operator is generating original content.

Google responded with two arguments. First, it said AI Overviews are “designed to reflect the information that exists on the web.” Second, it pointed out that users “can dig deeper and verify” the claims. The court rejected both. Framing a fabricated claim as a reflection of existing web content does not make it so. Offering a verification path does not transfer legal responsibility to the reader.

For publishers and brands, this ruling reframes the problem entirely. A defamatory AI Overview has, until now, occupied a legal gray zone: frustrating, potentially damaging, but difficult to litigate because Google could argue it was merely aggregating signals. Munich’s court has closed that gap, at least under German law. An AI Overview that makes a false and reputation-harming claim is now, in that court’s reading, a statement Google made.

Generative engine optimization (GEO) is the practice of structuring content and entity signals so that large language models surface accurate descriptions of a brand or publisher in AI-driven answers. GEO practitioners have focused mostly on visibility. The Munich ruling adds a second priority: accuracy monitoring with legal awareness.

The practical implication is direct. Brands and publishers should be running regular searches on their own entity names in markets where AI Overviews appear and documenting what the Overview claims. If those claims are false and reputation-harming, the Munich ruling suggests a named liable party now exists. That changes the calculus on when to engage legal counsel, and it changes the urgency of GEO corrections submitted through Google’s Search Console feedback and official dispute channels.

Google’s core defense, that its product is a neutral mirror of the web, has now been tested in an adversarial proceeding and found insufficient. The court’s ruling is specific to Germany and to the facts of this case. It will not automatically govern outcomes in the United States, the UK, or elsewhere. But it gives legal teams in other jurisdictions a concrete precedent to cite, and it signals to platform litigants that the “we just surface content” argument is not unassailable.

The immediate action for any brand with significant search presence is this: audit your AI Overview representation today, document false claims with timestamps, and treat any defamatory Overview as potentially actionable rather than dismissible.

Reported by The Decoder (the-decoder.com), published June 9, 2026, and updated June 11, 2026, covering the Regional Court of Munich ruling in case 26 O 869/26.