Lawsuit against Google AI: False search information can justify injunction

The Frankfurt Regional Court has clarified that AI errors can be classified as potential impediments to competition under certain conditions. This opens up the possibility for companies to defend themselves against false AI texts based on antitrust law. The background to this development is the growing concern about the “zero-click” effect, where users find their answers directly in Google's AI summaries and no longer visit the websites of the original creators.

The urgent proceedings concerned the depiction of a medical procedure for penis enlargement (Case No.: 2-06 O 271/25). A group of doctors criticized an AI overview that falsely claimed that a hidden part of the penis is severed and relocated outwards during the procedure. The plaintiffs saw this as a business impairment due to declining click numbers and attempted to have Google prohibited from distributing it, citing antitrust law and the Digital Markets Act (DMA).

Although the application failed in this specific case, the judgment published on September 10th contains signals for companies that depend on search engine traffic. For example, the judges found that German courts have international jurisdiction and that local law is applicable. Since the Federal Cartel Office has already certified Google's cross-market significance, the company is subject to special abuse supervision.

Google Got Lucky

Affected parties can therefore generally invoke the antitrust prohibition of abuse according to Sections 33 and 19 of the Act Against Restraints of Competition (GWB), reformed in 2021. The court explicitly considered it possible that objectively false information in AI overviews could constitute an unreasonable impediment to competition. It sees a significant risk potential for the common good, especially with health-related information.

However, one of the central legal questions remained unresolved for now: the 6th Civil Chamber left open whether Google should attribute the AI texts as its statements or whether it is merely an aggregation of third-party information.

The reason the doctors' application was unsuccessful was the high legal hurdle of “unreasonableness.” The court emphasized that a liability for injunctive relief only exists if there is an unreasonable impediment. High requirements must be met for this, which are not fulfilled with every inaccuracy. In a comprehensive assessment, the chamber concluded: The false statement, in its specific overall context, was not so serious for the average user that it would have justified an immediate preliminary injunction. Google benefited from the judges' view that the error could be “healed” by the context.

Error Rate Kept Under Wraps

The allegations regarding the Digital Markets Act were also unsuccessful, as the judges classified the AI overview merely as part of the search results and not as a separate product within the meaning of Article 6 DMA. For IT lawyer Jens Ferner, this interpretation underscores that AI-generated summaries are not automatically considered independent services as long as they relate to the search query and do not displace external offerings. Nevertheless, law firm Plutte assesses the judgment as “good news”: because the regional court considers it fundamentally possible that false AI information can unreasonably impede other companies.

Google itself declined to comment on the court ruling when asked by heise online. The US company keeps official statistics on error rates under wraps. To minimize risks, Google has throttled the output rate for sensitive topics such as medicine or finance to sometimes less than one percent. Google also attempts to legally protect itself by including the note “experimental.”

Overall, according to experts, the ruling offers an initial orientation. However, it is not the final word on liability. It does not give Google a “free pass.” Search engine operator notices such as “AI answers may contain errors” are not sufficient if false information is systematically disseminated. While the ruling opens up the antitrust route for affected companies, it sets high hurdles, as the falsity must be demonstrable in the overall context and the impediment must be unreasonable. The success of future lawsuits will depend on whether other courts follow this search engine-friendly view or opt for a stricter attribution as their statement.

(mki)