German Court Blocks Google AI Overviews Over Defamation
Munich Court Halts Google AI Overviews Over Defamation Claims
Munich's Regional Court of Munich I (Landgericht München I) on July 2, 2026, issued a preliminary injunction against Google, ordering the company to stop generating false statements in its AI Overviews product that defamed two Munich-based publishing companies Security Boulevard. Case number 26 O 869/26 centered on AI hallucinations—false or misleading content generated by the system—that falsely associated the plaintiffs with subscription scams, fraudulent business practices, and deceptive commercial activities.
The court grounded its decision in German Civil Code sections § 1004 (injunction against interference with property and legal rights) and § 823 (tort liability for damages caused by unlawful acts) Lexology. The preliminary injunction means Google must immediately stop the AI system from generating those specific false statements about the plaintiffs' businesses. However, this is not a final liability judgment: the case remains ongoing, and Google announced plans to appeal the decision.
What Makes This Different: Preliminary vs. Final Ruling
A preliminary injunction is a temporary court order that halts harmful conduct while litigation proceeds. It does not determine fault or award damages. The Munich court found sufficient legal grounds and evidence of irreparable harm to warrant immediate action, but the full trial on liability and monetary damages has not yet occurred. Google's appellate strategy remains unclear, though the company's statement signaled resistance to the ruling.
This case highlights a growing tension in the AI agent economy: generative systems routinely produce false or misleading summaries of real businesses, and defamation law—originally written for human-generated content—is now being applied to algorithmic output. The plaintiffs' identities remain undisclosed in public filings, though court records confirm they operate in Munich.
Implications for AI Oversight
The ruling does not represent regulatory action by German authorities. It is a civil court case between private parties, not a government enforcement action. Nonetheless, it signals that German courts are willing to use existing tort law to compel AI companies to fix demonstrable harms. No sentences, fines, or dollar amounts were awarded in this preliminary stage—only the order to cease the defamatory AI outputs.
As AI agents expand into search, summarization, and commercial recommendations, defamation liability for hallucinations may become a material cost of doing business in jurisdictions with strong privacy and reputation protections.
