Comment by Hfuffzehn
5 days ago
If I get it correctly I like the ruling.
So Google has established a product called Search. For that product rules have been established. Google has monopolized that product.
Now Google is replacing that product with a new product. But they keep calling it the same thing. Because they want to keep their monopoly.
That is what has been deemed illegal. Gemini is not illegal. Pretending the worst version of Gemini is Search is illegal, because it breaks the rules established for Search.
But IANAL.
It has nothing to do with monopolies. Google was protected from defamation law with search because the page title and snippets were direct quotes from the linked result page. Whereas with AI overviews, the copy is written by a Google-controlled LLM.
It should be noted that defamation law has a very low threshold in Germany, to the point that businesses routinely sue Google Maps users for less-than-5-stars reviews. Google had to change their display of reviews because of this.
Three stars review is taken down for "libel": https://news.ycombinator.com/item?id=44734895
Google maps german policy: https://support.google.com/contributionpolicy/answer/1699727...
Meanwhile, service in Germany is still rather poor (especially if you have children), but at least no one can complain!
You cannot get sued for a review just because it's lower than 5 stars. But of course if you write something like "I found a dead cockroach in my pizza", or "I hard that they don't clean the dishes enough" in a review without proof, that's defamation. And it doesn't matter if you give 1 or 5 stars with the review.
66 replies →
> HNer gets a legal threat after saying that he didn't like a doctor: https://news.ycombinator.com/item?id=44734895
This post is written by an LLM, and possibly a complete fabrication.
9 replies →
Damn, that makes public facing LLMs a horrible liability
4 replies →
Here's an article from outside the tech sector, which I am linking to as a corroboration since this is quite surprising for those of us unfamiliar with German law.
https://www.thelocal.de/20260504/how-a-google-maps-update-ex...
See also: Germany's leader filing hundreds of criminal complaints against people who insult him on social media:
https://rmx.news/article/germany-chancellor-merz-quietly-fil...
5 replies →
That's BS.
Google does not get routinely sued.
Yes, Lawyers send takedown requests to Google.
Google sends the reviewer a message and ask for a statement. If you provide one. They will check it and decide whether they seem it defamation or not.
However, Google doesn't verify shit. Even if you send them proof of a purchase or visit and your message is objectively a opinion (not defamation) they will take it down. Why? I assume because they really don't care about individual reviews and rather spare the time and money and just take the comment down.
Google changing their review displays is Google's decision. It has NOTHING to do with any threat of being sued.
They could actually leave the review up when it's not defamatory but they simply don't want to invest the time to validate it or risk leaving a potentially defamatory review online.
And the legal threat is without proof, so why would you even bring it up. OP could have simply posted the letter. They didn't wich makes this completely unprovable
the threshold is not low, I would argue, it's just companies like to exploit it.
(I just wrote an extensive analysis about that https://nickyreinert.de/2026/2026-05-28-google-maps-reviews/
Yes, the monopoly is not relevant for the court.
It is relevant for Google though, because they want to transfer it to another product.
And the court is saying that whatever that new product is, Google is not allowed to mislead the public by pretending it is search.
I'm not sure. Is it really just the misleading part that the court takes issue with, i.e portraying Gemini output as a search result?
In my view, the ruling could mean that Gemini's output is legally seen as first-person speech by Google regardless of where it is published.
8 replies →
I recall "Large Libel Models" was one of the sobriquets going around.
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4546063
Brilliant
Right. If Google isn't liable for that content nobody is.
That's probably what they actually think at Google. A lot of the "they're just like people" marketing now seems like responsibility-washing. That despite Google developing the LLM and charging for its use, they treat it less like a piece of software and more like providing access to some guy's answers who has no agency and isn't owned by them. If their software ever messes up and calls you a murderer to a potential employer, or offers dangerous advice, their reaction would be "oh well, that's just too bad isn't it? Well, we wrote that the LLM may not always be right in page 683 of the service agreement, so this is not our problem."
Making the response consist entirely of direct quotes sounds like a better user experience than what they're doing now.
So if the actual page is defamatory then Google quoting it verbatim is protected.
But if Google accurately summarizes the defamatory page, then the summary is defamatory?
It could presumably accurately summarise it while still sticking to reported speech. In German there is a grammatical tense that is basically only used for this; it's pops up in newspaper reporting a lot where they don't want to endorse the statements they are reporting.
Did you read the article?
The whole point is that these particular defamatory claims weren't made by some third-party, but hallucinated entirely by Google's own AI.
There is no "accurate summary". Just a pack of lies fabricated in-house at Google.
So AI isnt just ripping off the underlying information?
That's not what this article says. If you think the article mispresents the ruling, then you should point it out. You say you like the ruling then proceed to listing things that you like, but are not related to the ruling.
> If you think the article mispresents the ruling, then you should pint it out.
It seems that's exactly what they did.
Where? Honestly can't find it. Or you're making a joke about my typo?
I dont really understand your reasoning at all, nor what law it would break. What law do you think it would break?
It doesnt even seem to be what the article is saying. This looks like a section 230 sort of issue. Section 230 is a US law that protects platforms like facebook, google etc. being treated as publishers because the information, presumably, is just being passed through. But Germany is saying the AI results are authored by Google.
I don't see a world where AI results aren't reasonably considered the output of the company. They're minced and sausagified regurgitations but they're not the original sources either.
Yeah, but my point is: what does that have to do with monopolies changing their services?
And to continue your thought, what does that imply about copyright of training data? If Google is authoring the output it seems harder to argue they are ripping someone else off.
It really seems like a tightrope to say google is publishing their own opinion but their opinion is also just someone else's work.
1 reply →
Like it or not, this is how people perceive output from llms. Those ultra rich companies can't forever have the cake and eat it too, just milking global markets forever (I know they of course desperately want and need to, but that's their struggle).
I am fine with that solution since I don't need to shill for them for some ie investment or employment reasons. Less technically skilled people (aka your parents or grandparents) are getting their lives fucked up left and right because they learned to trust search results, and now they suddenly can't.
Own. Your. Shit.
The court ruled that the AI generated content has an author/editor/publisher: Google. It also ruled that Google can be held liable. Insert pikachu face meme.
[flagged]
None of that is in the ruling.
https://the-decoder.com/wp-content/uploads/2026/06/26_O_869_...
"Tatbestand
Die Verfügungsklägerinnen begehren von der Verfügungsbeklagten die Unterlassung von Darstellungen KI-generierter Antworten in einer Suchmaschine.
...
Die Verfügungsbeklagte betreibt eine Internet-Suchmaschine unter google.de in Deutschland. Dabei werden nach der Eingabe von Suchbegriffen durch eine die Suchmaschine benutzende Person auf den „Suchbefehl“ hin mithilfe bestimmter Algorithmen nach möglicher Relevanz sortierte Ergebnisse angezeigt. Zusätzlich bietet die Suchmaschine auch als unterstützende Funktion ein Suchergebnisformat an, bei dem mithilfe einer generativen künstlichen Intelligenz (im Folgenden:KI) repräsentative Ergebnisse zusammengefasst und angezeigt werden."
No "Suchmaschine" + no „Suchbefehl“ -> no "Tatbestand"
No "search engine" + no "search command" -> no "elements of a crime"
> No "Suchmaschine" + no „Suchbefehl“ -> no "Tatbestand"
I think you have misapprehended the ruling. Of course running an LLM that spits out libel is also a crime outside of search engines. It's just that _this_ crime has those elements. If Meta's LLM defamed a business, then the ruling would instead talk about Whatsapp users and chats, or whatever.
1 reply →
Which says nothing at all that is like your claim.
I mean was google search responsible for surfacing correct answers? I am not a defend of AI but this seems like a weird piece of legislation.
Original search before Google started trying to provide their own answers was purely pointing to relevant pages, even the first iterations of the first results being replaced by the result Google believed provided the correct answer could be pointed to as simply providing an answer someone else wrote (and to my knowledge was mostly fact based questions; birth dates, etc that are hard to categorize as defamatory). Now Gemini is combining and mixing together multiple sources to provide a new amalgam answer that IMO is distinct enough, and applied to touchier subjects importantly because they're treating the search bar like you're talking directly to Gemini, that it crosses a line between referencing speech by other people without endorsing (OG Search) and having the company produce speech about the search (new Gemini infested Search).
No, Google search was simply repeating quotes from other websites - those other websites would be responsible for their content. Now that Google is manufacturing answers using their own LLM, they are responsible for whatever results that produces.
Were the websites actually held responsible though?
> Now that Google is manufacturing answers
There isn't someone at a keyboard typing content. It's still just search, repeating quotes from other websites. The difference is that the algorithms have turned lossy, so the quotes are not always preserved in their original form.
5 replies →
> was google search responsible for surfacing correct answers
No, as the article lays out, google hallucinated facts that werent present in source material and so at that point the court found them liable for that.
Previous protections existed for search because it has been argued (and agreed) that they are merely a vessel for showing the content to the user. But when they begin to editorialize and reword the content to show their own version of the content (the AI summary) those protections dont apply even if the AI summary is shown in the same UI as search.