A German court has issued a preliminary ruling that Google can be held liable for false statements made in its AI Overviews. The case arose after two publishers were labeled as scams by the tool; the publishers sent Google a cease-and-desist letter, which the court said Google ignored.
The court reasoned that AI Overviews do more than present a list of links. According to the judgment, these summaries make “independent, new, substantive statements,” and therefore Google can be prohibited from repeating them. The court issued an injunction preventing Google from continuing to publish the disputed assertions.
The ruling is notable because it may be the first judgment anywhere to treat an AI company as responsible for the speech of its system. Traditional search engines have enjoyed a degree of legal protection on the basis that they merely surface other people’s speech via links; the court found that protection inapplicable to AI-generated summaries in this instance.
The court also rejected the industry assumption that liability can be shifted to users through a disclaimer. While surfacing links has been treated as an unavoidable, technical function, AI summaries are an optional feature — and therefore, the court held, they do not automatically qualify for the same legal shield, especially when the AI invents claims that do not appear in any source.
The practical stakes are significant: the reporting cited in the case notes that Google’s Gemini overviews provide incorrect source links 56% of the time. If such behavior is widespread, it could result in millions of false statements being presented to users each day.
In short, the court’s preliminary decision challenges the notion that search engines or AI tools can escape responsibility via warnings to users, and underscores that when AI systems generate confident, new assertions, the operators behind those systems can face legal consequences.
