Category: News | Published: 2026-06-09
For roughly three decades, search engines have operated on a legal principle that made the modern web possible. If a search engine links to a page containing false or defamatory content, responsibility for that content lies with whoever published the page, not the search engine that helped people find it. Search engines are intermediaries. They point. They do not speak.
A ruling from a German court in May 2026 has challenged that principle in a way that could reshape how AI-powered search is treated under the law. The court decided that when Google's AI search feature generates its own answer, Google is no longer pointing. It is speaking. And if it says something false, it bears responsibility for having said it.
What Actually Happened
The case was heard by the Landgericht München I, the Munich Regional Court, under case reference 26 O 869/26. The preliminary injunction was issued on 28 May 2026, following an oral hearing in April of that year.
Two publishing companies that are part of a Munich-based media group discovered that Google's AI Overviews feature, the AI-generated summaries that appear at the top of certain Google search results, had been falsely associating them with scams, subscription traps, and questionable business practices. The descriptions were serious and the kind that could deter customers, partners, or advertisers.
The publishers first sent cease-and-desist letters to Google. Google did not act on them. The publishers then took the matter to court.
What the AI Actually Said
The detail that makes this case particularly striking is that the false statements were not sourced from anywhere. The AI Overviews feature had generated language along the lines of stating directly that the company in question was known for dubious business practices. That phrasing did not appear in any of the sources that the AI cited. The AI had not summarised a negative article about the company. It had created a negative characterisation in its own voice, with no factual basis in the linked material.
This matters enormously to the legal analysis. A search engine linking to a critical article is an intermediary. An AI that invents a critical characterisation and presents it as settled fact is doing something categorically different.
Why This Breaks 30 Years of Legal Precedent
For most of the search engine era, courts across Europe and the United States have afforded platforms significant protection from liability for third-party content. In Germany, the Federal Court of Justice had established that traditional search engines carry limited liability precisely because they make third-party content findable rather than creating content themselves. A proactive duty to check every search result for accuracy would make search engines unworkable.
The Munich court examined that reasoning carefully and concluded it simply does not translate to AI-generated search results.
The court drew the distinction clearly. A regular search engine points to what others have written. Google's AI search feature synthesises information from multiple sources, rewrites and restructures it, and presents the result in its own voice as a direct answer to a user's question. That is not pointing. That is publishing.
The judges concluded that AI Overviews create what they described as independent, new, and substantive statements. Google controls the AI model and the algorithms that produce those statements. Google therefore carries responsibility for them in the same way a publisher carries responsibility for what it prints.
Why Google's Defence Did Not Succeed
Google's primary argument was that users have access to the sources linked within AI Overviews and can therefore verify any information for themselves. The court rejected this.
The judges drew an analogy to headlines or short teaser text. In practice, many people read a headline and absorb its message without reading the full article beneath it. The same behaviour applies to AI-generated answers: if the answer appears complete and authoritative, a significant proportion of users will take it at face value without investigating further. The existence of cited sources does not neutralise the impact of a false statement presented as a confident conclusion.
The court also addressed whether AI Overviews are essential to the functioning of search in the way that traditional results are. It concluded they are not. Users can find information through ordinary search links without an AI-generated summary. That distinction allowed the judges to apply a higher standard of responsibility to AI-generated content than has historically applied to the links beneath it.
The Penalty Google Faces
The injunction prohibits Google from repeating the specific false claims about the two publishers. Breach of that prohibition carries potential fines of up to 250,000 euros per violation, or custodial sanctions in serious cases. Google was also ordered to bear 80 per cent of the court and legal costs.
These are not trivial consequences for a single preliminary injunction, and they signal that German courts are prepared to treat AI-generated defamation as a serious legal matter rather than an incidental by-product of a technical system.
Google Is Appealing
Google has confirmed it intends to challenge the ruling. A company spokesperson told Reuters that the case concerns specific errors rather than the fundamental operation of Google AI search features, and that the company disagrees with the court's conclusions. Google also noted that the overwhelming majority of AI Overviews are accurate and that it takes action when policy violations are identified.
This is an important caveat. The Munich ruling is a preliminary injunction, not a final appellate decision. The legal position could still change as the case progresses through the German court system. However, the reasoning set out in the initial judgment is detailed and considered, and it represents a genuine challenge to the legal framework that has protected search platforms for decades.
The Wider Implications for AI-Powered Search
The Munich ruling matters well beyond this particular case because of the principle it establishes. If an AI system does not simply retrieve and link to third-party content, but instead synthesises that content and presents its own conclusions in its own voice, the intermediary shield that search engines have relied on does not apply.
That principle, if it holds on appeal and is adopted by courts in other jurisdictions, has significant implications for Google AI search specifically and for AI-generated content more broadly. Enterprise AI assistants that answer questions by synthesising company documents and presenting confident summaries, customer service bots that generate responses based on product information, chatbots that combine sources and draw conclusions: all of these involve the same basic dynamic as AI Overviews. An AI creates an original statement. If that statement is false and causes harm, who is responsible?
The Munich court's answer is that the company providing the AI is responsible. That answer is not yet settled law across Europe or globally. But it is now on the table in a way it was not before.
What This Means for Businesses Using AI
For businesses deploying AI tools that generate content, answers, or summaries, this ruling is a prompt to think carefully about oversight and accountability. AI systems that synthesise information and present confident answers can be wrong. When those wrong answers concern real companies or real people, the legal and reputational consequences may rest with whoever deployed the AI, not whoever provided the underlying data.
Good AI governance includes understanding what your AI tools are generating and having processes in place to catch and correct errors before they cause damage. That is as true for a customer-facing AI assistant as it is for Google AI search at global scale.
If you are thinking about how to adopt AI tools in your business responsibly, including how to evaluate the risks and put appropriate oversight in place, our AI Consultancy page is a useful starting point.