Bavarian court – A Bavarian civil court ruled that Google can be held liable for hallucinations produced in its search AI summaries, rejecting the argument that the underlying facts must have existed somewhere on the internet. The decision is not final and Google is expected t
In a Bavarian courtroom, the fight wasn’t about whether an AI could generate convincing text. It was about who pays when that text harms real companies.
The ruling came in a civil case brought by two Munich companies. Both said they were wrongfully slandered by large language model hallucinations shown in Google’s search results—specifically. the AI-generated summary text that appears at the top. Google’s defense leaned on a simple idea: the information must have existed somewhere. and the company shouldn’t be responsible for what that summary claims—almost as if it were just routing users to harmful material.
The judges didn’t accept that comparison. They treated the AI-generated content as something Google itself creates, not something Google merely links to. The court’s view was that this was text generated by Google’s system. not a web page presented through links. So when the summary states something defamatory and false, liability can attach to the creator of the text.
Google also argued that if the court was looking for a way to limit responsibility, the mechanism mattered. But the ruling made that point irrelevant. The judges’ reasoning held that the quality—or the “model” doing the generation—doesn’t change what the company is doing. Google is producing the summary. and if that production produces slanderous claims. it can be held accountable the same way it would if a human employee wrote the text.
The case carries obvious stakes for anyone who trusts those top-of-results summaries. If the decision stands on appeal, it would set a high bar for how search AI systems are deployed in the Federal Republic.
The ruling isn’t final. Google has the right to appeal, and it is expected to do so. Still, the pressure is clear: if the appeal fails, it’s not hard to imagine Google changing course—possibly by pulling the AI summaries from searches in Germany.
For now, the takeaway from the courtroom is straightforward. When an AI summary harms people on the other side of the screen. the person responsible won’t be allowed to hide behind the idea that the truth must exist “somewhere.” In the judges’ view. the summary itself—generated by Google—counts as something created. and therefore something that can be judged.
Bavarian court Google Gemini AI summaries search hallucinations legal liability Munich companies Germany appeal
So basically Google is “making it up” and now they have to pay? Seems fair I guess.
I don’t get it, if the facts are already out there on the internet then how is it Google’s fault. Like the AI is just summarizing, right? But they said it’s created by Google so idk.
Wait, the headline says “Real Boy” fiction?? I thought that was about some movie or scandal. I skimmed and now I’m seeing “hallucinations” and “defamation” like… so the summary said something true but in a weird order? Courts love making tech harder.
If they’re gonna hold Google liable for AI summaries, then they should just turn that whole feature off. I mean Germany is probably gonna force it, and then everyone else will copy. Also what if the AI summary is technically accurate but still “hurts” some company’s feelings, does that count too?