A German court finds Google liable for what its AI says in search overview

A German court has found Google liable for what its AI overview says in Google Search. Google was given an injunction in the case where two publishers found that AI overview gave false information about them. The court says that AI overview is unlike traditional search, as it provides information in its own words and structure.

In Short

• Two Munich publishers said AI summaries wrongly tied them to scams
• Judges said the feature creates original statements beyond linked source material
• Google's claim that users could verify answers themselves was rejected

Google’s push for AI tools spans across its entire ecosystem, including Google Search. During I/O 2026, the company announced that Search was going to get its biggest change in almost 25 years, as AI overviews become more prominent on-screen. But Google has faced a setback regarding AI overviews in Germany, where a court has found the company liable for what the AI says.

The Regional Court of Munich has held Google directly liable for false statements generated by its AI overviews in Search. The court issued a temporary injunction against Google in a case brought by two Munich-based publishers who alleged that the AI overviews were generating false claims about them.

Do note that Google's AI overview feature has faced criticism in the past for seemingly being broken when users asked it to search for definitions of words like "disregard."

What was the case?

The case arose after Google’s AI overviews allegedly linked the two publishers to scams, subscription traps, dubious companies, shady business practices, phone calls that never happened and claims of non-availability.

According to the court, the AI had mixed up information about other companies and created connections that did not appear in any of the linked sources. The publishers first sent Google a cease-and-desist letter, but moved to court after what they said was an inadequate response.

In its ruling, the court classified Google as a direct infringer because the AI Overview was its own content rather than a simple list of search results. It said the AI rewrites and assesses results “in its own words and according to its own structure.”

In one example cited in the proceedings, the overview opened with the statement, “Yes, [company] is known for dubious business practices,” and then set out a summary, alleged warning signs and tips for users.

AI overviews not like traditional search results

While search engines have limited liability for traditional results since they only direct users to different links as per previous court rulings in Germany, the court ruled that things were different for AI overviews. Since, as per the court, the feature produces Google’s own content by generating “independent, new, and substantive statements” after evaluating and combining material from various external sources.

The court stated that some of the claims in the overview were “not even made in the search results”. It therefore treated those statements as “the defendant’s own statements”. The court further said Google owns the material produced by its AI “because it alone has influence over the AI’s offering and the algorithms with which the AI operates.”

As per the court, AI overview is “by no means absolutely necessary” for using the internet and is an additional feature beyond ordinary search results.

Court dismisses Google’s argument on AI overview

During the hearing, Google argued that users could verify the AI summary by checking the linked sources and that users generally knew “that information generated with AI should not be blindly trusted.”

But the court rejected that defence, saying the fact that a statement can be disproved through further research does not “regularly exempt from liability for this statement.”

It said the AI Overview was “understandable on its own” and amounted to “a self-contained statement with independently understandable content and no reference to other possible interpretations or even unreliable content”. The ruling also noted research cited in the case, including Pew findings that users are much less likely to click on sources shown through an AI Overview.

The court drew a comparison with press law, saying publishers can be liable for stand-alone teasers even if readers never open the full article. It also pointed to what it described as a protection gap – if Google were liable only for obvious violations, people targeted by false AI statements would have little effective remedy, especially where the third-party websites used as sources had not made those statements themselves.

On that basis, the court said Google could not rely on host provider protections under the Digital Services Act or on the usual notice-and-take-down framework applied to search engines.

This ruling may have international reach, says court

The court granted most of the publishers’ claims and held that the risk of repetition remained even though the specific wording was no longer being displayed, because Google had not issued a cease-and-desist declaration carrying a penalty and the algorithms could generate the same statements again. Google was ordered to bear 80 per cent of the legal costs, with the two publishers paying 10 per cent each.

The court also said the ruling could have international reach. This could potentially make Google liable for its AI overviews globally.

The case comes amid wider scrutiny of AI-generated search responses. As per a report from the New York Times, Google’s AI overviews using Gemini 3 answered correctly 91 per cent of the time, while 56 per cent of correct answers could not be backed by the linked sources. The Munich court’s ruling addressed that same issue directly, holding that when AI overviews make claims that do not appear in the cited material, Google can be held responsible for them.

- Ends