The Regional Court of Munich has hit Google with an injunction (case 26 O 869/26) over AI Overviews that falsely tied two Munich publishers to scams and subscription traps, ruling the company directly liable for what the feature writes.
The court rejected the search-engine liability shield that normally protects a list of links. An AI Overview, it found, rewrites and judges results "in its own words and according to its own structure," opening with confident lines like "Yes, [company] is known for dubious business practices" and even making claims that appear in none of the linked sources. Because Google alone controls the model and its algorithms, the court called the output "the defendant's own statements."
If higher German courts uphold the reasoning, every confident hallucination in an AI Overview becomes a potential defamation claim against the platform, not the cited sites. That is a different risk calculus for anyone shipping generative answers in Europe.