Munich court finds Google directly liable for AI Overview misinformation
A German judge has issued a temporary injunction against Google after its AI Overviews platform spread false claims about two local publishers, marking a significant shift in how AI-generated search features are treated under liability law.
The Regional Court of Munich has ruled that Google is directly liable for incorrect information presented by its AI Overviews platform, classifying the service as the company’s own content rather than neutral search results. The decision follows a case where the AI algorithm conflated two separate Munich-based publishers, inventing facts about one entity that were not supported by the cited sources. The court issued a temporary injunction prohibiting Google from spreading such misinformation, rejecting the tech giant’s argument that users should verify linked sources.
Under German law, search engine operators are generally protected from liability for third-party content indexed in search results. However, the court determined that AI Overviews create "independent, new and substantive statements" by evaluating and combining content from third-party websites. Because the AI rewrites information in its own words and structure, the court classified Google as a direct infringer, meaning the standard liability protections for search engine operators do not apply.
The legal action stemmed from an incident where the AI confidently suggested one of the publishers was known for dubious business practices, including scams and subscription traps. The algorithm drew connections that did not appear in any linked sources and did not actually exist, effectively pulling information about a different entity. The publishers sent a cease-and-desist letter, but stated Google did not respond appropriately, leading to the court’s intervention.
Google argued at the hearing that users are aware that AI-generated information should not be blindly trusted and can verify claims via source links. The court rejected this defence, noting that research indicates the presence of AI summaries significantly reduces the likelihood of users clicking on source links. A study by Pew Research cited in the proceedings found that just one percent of users click on source links after reading an AI overview, undermining the argument that users can easily verify the accuracy of the summaries.
Recent studies highlight the scale of the accuracy issues within the platform. A report noted by The New York Times suggested that AI Overviews get facts wrong around nine percent of the time. With Google reporting that 2 billion people interact with the feature monthly, this translates to a high volume of incorrect queries annually. Furthermore, an analysis found that 56 percent of correct answers could not be backed up by the linked source, leaving users with little way to check the AI’s work.
Users who wish to avoid AI Overviews can opt out by appending "-ai" to their search queries. While the exact long-term impact of this ruling on other jurisdictions remains unclear, the decision establishes a precedent that AI-generated summaries may be treated as direct content creation, carrying full responsibility for factual errors and defamation.
