Berlin vs. Munich: German courts split on Google AI liability

Key Takeaways

• Berlin court ruled AI Overviews are just a new search format, not original content requiring publisher liability
• Munich court reached the opposite conclusion days earlier, holding Google liable for AI-generated false claims
• The conflicting rulings create legal uncertainty that could affect every company building AI search products

A Berlin court has ruled that Google's AI Overviews are simply a new search result format, not original content that would make the company liable as a publisher. The June 2026 decision directly contradicts a Munich court ruling from days earlier that held Google responsible for false claims generated by its AI. Two German courts, one week apart, reached opposite conclusions on the same fundamental question: when does a search engine become a publisher?

The Berlin case started with a perfume company lawsuit. When users searched for fragrance imitations, Google's AI mentioned the company's protected brand names alongside cheaper knockoffs and linked to websites selling alternatives. The company argued this constituted trademark infringement.

The court dismissed the case. Its reasoning: the AI summary is just aggregating information already available on third-party websites. An average user, the judges argued, would recognize that Google is pulling together external content rather than making its own claims. The search engine has no "decisive influence" over what the AI says.

“The AI summary is a new format for summarizing third-party content and the provider has no decisive influence over the output.”

— Berlin Regional Court, Official Statement

Why did Munich reach the opposite conclusion?

The Munich case involved a different set of facts. Google's AI had falsely connected two publishers to fraudulent schemes. These claims did not appear in any of the sources the AI cited. The court found that Google's AI was generating novel statements, not summarizing existing content.

The Munich judges rejected the argument that users could simply verify information themselves. If nobody bears responsibility for AI-generated falsehoods, victims of defamatory or inaccurate AI claims have no legal recourse. That gap concerned the court enough to treat Google as a publisher.

What's the real difference between these cases?

The legal foundations differ, which explains some of the divergence. Berlin dealt with trademark and competition law. The AI had accurately summarized what third-party websites said about perfume knockoffs. Munich dealt with defamation. The AI invented connections that existed in no source material. Accuracy versus hallucination.

But the core disagreement runs deeper. Berlin says Google is a neutral intermediary displaying external information in a new format. Munich says Google transforms information through AI processing and must answer for the result. Both cannot be correct.

Does the Berlin ruling hold up to scrutiny?

The Berlin court's logic has gaps. The claim that Google has no "decisive influence" over AI responses seems difficult to defend. Google selects the AI model. Google sets the system parameters. Google decides how responses are structured and displayed. Every one of these choices shapes what users read.

There's also a behavioral problem. Research suggests most users treat AI summaries as complete answers and never click through to sources. They consume AI responses like editorial content, regardless of whether courts think the aggregation is obvious. The Berlin court assumed a level of user sophistication that may not exist.

And as Munich demonstrated, AI summaries sometimes contain claims that appear in no cited source. A summary that invents information is not a summary. It's generation.

What happens next for AI search liability?

Both cases will likely face appeals. Higher German courts will need to resolve the contradiction. The outcome matters beyond Google. Every company building AI models with internet access faces the same question: are you a search tool or a publisher?

The European Digital Services Act adds another layer of complexity. The DSA was written for traditional platforms and search engines. AI-generated snippets that sometimes hallucinate or misrepresent businesses fit awkwardly into existing categories. Regulators are watching these cases closely.

If the Munich interpretation prevails, AI search providers will need robust fact-checking mechanisms or face liability for every error. If Berlin wins, companies harmed by AI hallucinations may have no legal remedy against the platforms that generated them.

The business stakes are enormous

Google has invested heavily in AI Overviews as the future of search. Perplexity, Microsoft Copilot, and dozens of startups are building similar products. The entire AI search category depends on being able to synthesize information without assuming publisher liability for every claim.

But businesses featured inaccurately in AI summaries have legitimate grievances. A perfume company watching its brand name appear alongside knockoff sellers suffers real harm. A publisher falsely linked to fraud suffers worse. Someone has to be accountable.

The appeals courts will need to draw a line. Right now, that line doesn't exist.

Frequently Asked Questions

Are Google AI Overviews considered original content under German law?

German courts are split. Berlin ruled they are just a new search format displaying third-party content. Munich ruled they are independent content making Google liable as a publisher. Appeals courts will need to resolve the contradiction.

Can you sue Google for false information in AI Overviews?

Under the Munich ruling, yes. The court held Google directly liable for AI-generated claims that did not appear in cited sources. Under the Berlin ruling, no. The court found Google is merely surfacing external information.

How does this affect other AI search companies?

Any company providing AI-generated summaries with internet access faces the same legal uncertainty. The eventual appeals court decisions could establish liability standards for the entire AI search industry across Europe.

What's the difference between the Berlin and Munich cases?

Berlin involved trademark claims where the AI accurately summarized third-party websites. Munich involved defamation where the AI invented false connections not present in any source. The different facts led to different legal reasoning.

Source: The Decoder / Matthias Bastian