German court rules Google liable for false statements in AI Overviews

At a glance:

• A Munich regional court preliminarily ruled Google liable for false statements generated by its AI Overviews feature.
• The court ordered Google to remove defamatory content and cover 80% of the legal costs from the case.
• The decision challenges the traditional immunity of search engines when they use generative AI to create new statements.

Background of the case

The lawsuit originated after two publishers discovered that Google’s AI Overviews linked their names to questionable business practices, scams and subscription‑related frauds without any basis. Earlier this year the publishers sent Google a cease‑and‑desist letter, which the company rejected, arguing that its summaries carry a disclaimer about possible errors and that users should verify information independently. The Munich Regional Court took up the matter and issued a preliminary ruling that treats the AI‑generated summaries as independent, new statements rather than mere links to third‑party content.

The court’s analysis found that Google’s AI combined data from other firms flagged for illicit practices with information about the plaintiffs, producing associations that never appeared in any of the source pages returned by the search engine. This synthesis was deemed a “substantial” creation of false claims, distinguishing the feature from conventional search results that merely display links.

Court's reasoning and implications

Judges held that, unlike traditional search engines that merely facilitate access to third‑party content, Google’s AI Overviews can generate nonexistent claims by recombining information from multiple sources. Because only Google can modify the underlying technology, the court concluded it must be held liable for the output. The judges also rejected Google’s reliance on the user‑verification warning, stating that such a disclaimer does not absolve the distributor of liability when the false statements never existed in the original sources.

The ruling further noted that AI‑generated results cannot be shielded by free‑speech protections, as they are the product of an algorithm designed, trained and managed by a corporation, not an individual’s expression. As a precaution, the court required Google to remove a large portion of the defamatory statements and to pay 80% of the legal costs incurred in the proceedings.

Reactions and next steps

A Google spokesperson, quoted by Ars Technica, said the company is reviewing the decision and may appeal, emphasizing its investment in the quality of AI Overviews and its aim to reflect information that exists on the web. The decision, while not yet final, could set a precedent for how courts treat generative AI features in search engines worldwide, potentially prompting other jurisdictions to examine similar liability questions.

Observers note that if the ruling stands, search engines may need to redesign AI summary tools to include stronger safeguards, tighter source attribution, or opt‑out mechanisms for publishers. The case also highlights the growing tension between AI‑driven content generation and existing intermediary liability frameworks, a debate likely to continue as more companies deploy similar technologies.