The Munich Regional Court ruled that Google is directly liable for false statements in its AI Overviews. Two German publishers won a preliminary injunction after the feature linked them to scams, subscription traps, and shady business practices. The court set a penalty of up to €250,000 per violation. Google bears 80% of court costs and says it will appeal.
Basically, the ruling says aggregating other people’s text is search, but generating new text is publishing. The search-engine liability shields that protect Google from being sued over what other websites say do not apply to the sentences Google’s AI writes on its own. The court specified three conditions: “independent, new, and substantive.” Once an AI produces independent, new, substantive statements, the operator owns them.
A study commissioned by The New York Times this spring measured 4,326 Google searches against the SimpleQA benchmark. The analysis found that while AI Overviews were approximately 91% accurate, source attribution was far less reliable. More than half of the cited sources did not fully support the claims being made. At Google’s scale, even a single-digit error rate could translate into an enormous number of incorrect AI-generated answers.
If this legal logic travels, operators of any AI that publishes a defamatory sentence may become defendants.
Every company needs a Claw strategy. Do you have one?
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various generative AI models.