On June 12, 2026, the Munich District Court issued a preliminary ruling holding Google directly liable for false claims generated by its AI Overviews feature in search results. The case was brought by two Munich-based publishing companies, including The Decoder, whose reputations were damaged after being inaccurately linked to fraud and unethical practices.
The court classified AI Overviews as Google’s own authored material rather than neutral links to third-party content. Unlike traditional search results, where the platform merely indexes and displays external publications, AI-generated summaries create entirely new text by interpreting sources. This fundamental distinction served as the basis for denying the platform the legal immunity typically granted to search engines.
Google cited its user disclaimers regarding the need for independent verification and argued that the feature functions similarly to standard search. The court rejected these arguments, stating that warnings do not absolve a company of responsibility when the system generates original content instead of simply aggregating existing data. The company intends to appeal the decision, maintaining that the dispute concerns specific errors rather than the underlying principles of the Overviews feature.
The ruling is based on an analysis of the generation mechanism: AI does not merely summarize; it paraphrases and structures information in its own words to create new assertions. The case specifically highlighted instances where Overviews falsely accused publishers of involvement in subscription scams and fraud without any basis in the source materials. The court noted that because only Google has the power to adjust the algorithm and its outputs, the company must bear full responsibility.
This verdict diverges from previous legal precedents in the EU and the US, where search engines have been viewed primarily as neutral intermediaries. Generative features now elevate the platform to the status of a direct publisher of information. Similar logic could be applied to other systems—ranging from OpenAI’s chatbots to Anthropic’s tools—where users also receive generated answers accompanied by verification disclaimers.
The industry will likely be forced to overhaul system architectures and disclosure policies as a result. Developers will either need to strengthen real-time verification and source-citation mechanisms or accept the risk of lawsuits over hallucinations and inaccuracies. While the preliminary nature of the ruling leaves room for appeal, it already signals a shift in the legal landscape: generative AI is moving from being a mere "tool" to a product for which its owner is legally accountable.
Questions remain regarding how this will scale: it is unclear if the precedent will extend to other jurisdictions and different types of AI applications, such as corporate chatbots and assistants. Independent audits and future litigation will determine how well the legal distinction between "search" and "generation" holds up over time.
The Munich court’s decision underscores that when creating new textual claims, a developer cannot shift the blame onto the algorithm or the user.
