A German court has ruled that Google is responsible for false information generated by its AI Overviews feature, a decision that could set a precedent for AI-powered search engines globally. The Munich Regional Court mandated that Google must prevent its search engine from spreading inaccurate claims.
This ruling stems from a case where two publishers accused Google's AI summaries of wrongly associating them with scams and fraudulent practices. The publishers sent Google a cease-and-desist letter after discovering these baseless accusations in search results.
Google had argued that its AI Overviews include a disclaimer warning users that information might be inaccurate and should be independently verified. However, the court found that Google's AI combined information from various sources, including those flagged for illicit activities, to create "independent, new, and substantial statements" that did not accurately reflect the original sources.
The court distinguished AI Overviews from traditional search engines, stating that while search engines merely link to third-party content, Google's AI actively generates new statements. This generative capability, the court reasoned, makes Google accountable for the accuracy of the content produced by its technology, as traditional third-party protections no longer apply.
Furthermore, the judges determined that Google's warnings about potential AI "hallucinations" do not absolve the company of liability. They argued that victims of false AI-generated statements would have no recourse if tech companies could simply deflect responsibility through user warnings. The court also concluded that AI-generated content, being the product of an algorithm designed and managed by a company, is not protected under free speech principles.
As a precautionary measure, the ruling ordered Google to remove a significant portion of the defamatory statements and cover 80 percent of the legal costs. A Google spokesperson indicated the decision might be appealed, stating the company is reviewing the ruling, which is not yet final.
This decision could have far-reaching implications for other AI companies like OpenAI and Anthropic, as it suggests that general disclaimers may not be enough to escape liability for AI-generated misinformation. The core argument is that when an AI creates new assertions not directly present in its training data, the company behind the AI bears legal responsibility for any resulting harm.
