Nobody needs AI to search the Internet, court says in ruling against Google

• "In the past, AI firms have hoped that disclaimers warning about misinformation would protect them from lawsuits over untrustworthy outputs. Last year, one chatbot maker even argued that AI speech is its own category of “pure speech” and the First Amendment should protect it."
  
  

• AI is not necessary to search the web
  
  Historically, any potentially harmful content surfaced by search engines has been protected from direct liability because that surfacing was considered largely unavoidable when helping users sort through an enormous tangle of information online. But the German court emphasized that AI search engines do not enjoy those same protections because AI summaries merely provide “an additional function—one without which the use of the search engine would still be (and is) possible, and without which users are perfectly capable of finding results amidst the ‘flood of data.’”
  
  In other words, nobody needs AI to search the Internet, so AI firms can’t just let their tools attribute false claims to fake sources without assuming any liability.
  
  
  
• Update, June 11, 2026:
  
  Google has provided us with a statement on the ruling. The company says its AI overviews are designed to "reflect" information that already exists on the web.
  
  "We invest deeply in the quality of AI Overviews to ensure that the overwhelming majority of responses provide accurate information, and they are designed to reflect the information that exists on the web. We’re carefully reviewing this decision, which is not yet final," a Google spokesperson said.
  
  Google adds that AI overviews can occasionally miss context or misinterpret web content, just like traditional search results. But that's exactly where the Munich ruling disagrees. The court draws a line between AI overviews, which generate new content loosely based on sources, and traditional search results, which list sources with direct quotes. That distinction is what makes Google directly liable, according to the court. In its statement, Google also repeats the very argument the court dismissed, that "people can dig deeper and verify."

 

Funny thing is that the ai was trained from books, so you do not need the internet to search for informations either.

 

It seems like an unnecessary case. What purpose does it serve?

Mis/dis/mal/information is as old as time, as is discernment.

 

Yah, I am a bit sleepy right now, but im having trouble understanding this haha.

 

The point is to curtail the spread of hallucinated misinformation or the reinforcement of existing misinformation. Google's AI overview hallucinated "alternative facts" regarding some companies and their alleged malpractices. Obviously hurt their businesses, or so they claim anyway.

The preliminary ruling says that Google is liable for its AI overviews, as, unlike plain search results, AI overviews no longer present a (weighted but still) neutral index of what already exists on the net. Instead, AI overviews hallucinate new alt-facts into existence that are exceptionally difficult to correct/dispel.

 

Not gonna pretend that I didn't watched some You Tube video about this whole issue recently, in nutshell, it takes information from other sites and pretty much traffic of most pages went absurdly down because people just read summary and move on, then it spits questionable answers, also pretty much favors their own Reddit for most of it, so it's mostly stuff that got must upvotes which means absolutely nothing about actual credibility, just echo chamber of the given board. There's more, but I just can't remember.

 

Hmm , the timing seems impeccable here , especially after Claude Fable was denied to people. And the government trying to leverage the use of AI to gather information from people all around without letting anyone have access to it. Especially the said thing can be used to find and attack vulnerabilities on anything we use nowadays.

 

the point is they cant escape liability by hiding behind this argument. the ai is not just repeating information...its generating NEW "information". the ruling is saying tech companies are responsible for the things their ai makes up.

what is the point? well besides the obvious of holding companies accountable for the things they do, there is the issue of purposely disseminating false information for any number of nefarious reasons, and then hiding from liability/accountability.

 

I would venture, based on recent interactions with some people who shall remain nameless and ageless, that it serves to un-institutionalize complete fabrications by ai agents in an era where students (from Primary to Secondary, and increasingly, Tertiary tiers of education) rely less and less on books written by experts, professionals, and academics, and blindly on whatever first pops up on a google search page results.

In past years, search engines, including google, provided a list of links to existing pages, articles or content.

Recently, search engines - starting with google - have been fabricating said content, without any indication of or links to fact sourcing.

What the court is saying is, to put it in the Queen's English: "This is bullshit."

 

It protects people from an information warfare ruled by big players like google.
If you'd google your name and Googles AI sais 'that individual is most likely a miserable fck that molestes minors". What would you prefer? Google being liable for that output or would you be like "never mind, it's freedom of AI speech".
It is not the same with search results. Maybe the first hit says you are exactly like that, but the second and the third could call you mother Theresa.
PS: yes misinformation is as old as mankind. And the battle against that is just as old.
The solution to misinformation is not just giving up.

 

I haven't found this to be true. I have been using Gemini and ChatGPT for some time and both give links to where they have sourced each piece of information. I find Gemini to be a natural successor to the old Google search. It saves a lot of research, handles the drudgery of cross referencing and gives far better results. Better than all that, it presents it's results as an informed, cohesive response rather than just random snippets from different sources.

 

As I and others had previously suspected, many AI-related matters and issues are being resolved in court; if it were up to the tech giants, we would no longer have the rule of law, but rather the law of the jungle—or to put it another way, whoever has the most money and power simply punches the others in the face. A good ruling—thanks to the court in Germany.

 

1. This case is about Google and a specific plaintiff's case scenario - it's not about you, or what you found or didn't "to be true."

2. Good for you.

 

The court makes several distinctions and rulings here that raise some eyebrows and fundamental questions for me:

1. Who decides what constitutes core function and what constitutes an optional additional function? And does this heightened standard of liability generally apply to every optional product enhancement?
2. If a stochastic process strings words together without its own intent, is liability for statements even the correct legal category when weighted randomness is a fundamental component of the system?
3. What role does the prompt play in assessing liability if the input significantly shaped the output? (“Yes, [it] is known for dubious business practices and is often perceived as a scam”)
4. Could the generated output be reliably reproduced?
5. And finally, a more meta-level question: If there is widespread public awareness of AI hallucinations, to what extent can a third party actually be harmed by an AI-generated statement? And how is that measured? After all, personal responsibility and common sense also play a recurring role in other court decisions. Not in this one, it seems...

 

Last edited: Jun 27, 2026 at 4:53 PM

The court did not make a general ruling on AI; rather, it dealt with a specific case involving a plaintiff and a defendant. Naturally, the ruling must be based on applicable legislation, such as the Basic Law and similar statutes.

https://www-spiegel-de.translate.go...l=auto&_x_tr_tl=en&_x_tr_hl=de&_x_tr_pto=wapp

Spoiler: Google haftet für KI-Lügen in der Suche

Urteil in München

Google haftet für KI-Lügen in der Suche

Google zeigt Nutzern nicht nur eine Trefferliste mit Links an. Sondern immer öfter auch KI-generierte Zusammenfassungen. Was aber, wenn die KI irrt? Das Landgericht München hat in einem solchen Streitfall entschieden.

11.06.2026, 16.19 Uhr

Google muss unmittelbar für Aussagen seiner KI-Übersicht in der beliebten Suche haften. Das hat das Landgericht München in einem nun bekannt gewordenen Urteil entschieden.

Wer bei Google eine Suchanfrage eintippt, erhält mittlerweile nicht mehr nur eine unkommentierte Liste an Linktreffern aus dem Netz. Sondern kann auch einen mittels KI generierten Antworttext angezeigt bekommen. Google hat diese KI-Übersicht im vergangenen Jahr eingeführt. Sie wird mittlerweile bei vielen Suchanfragen über den klassischen Suchergebnislisten eingeblendet.

Das führte im strittigen Fall zu Problemen: Einem Verlag wurde in dem KI-Textabschnitt fälschlicherweise unlauteres Geschäftsgebaren unterstellt. Er klagte – und gewann. Das Gericht erließ eine einstweilige Verfügung, die Google mehrere Aussagen verbietet.
Prinzipiell sind Suchmaschinenanbieter von der Haftung weitgehend freigestellt. Sie müssen fremde Angebote nicht präventiv auf Rechtsverletzungen überprüfen, aber auf etwaige Beschwerden reagieren. Wie weit dieser Haftungsausschluss reicht, ist immer wieder Teil rechtlicher Auseinandersetzungen. So stritt die Ex-Frau des ehemaligen deutschen Bundespräsidenten Christian Wulff, Bettina Wulff, jahrelang mit Google über ehrverletzende Suchvorschläge zu ihrem Namen. Die Auseinandersetzung endete mit einem außergerichtlichen Vergleich.

Suchergebnisse stützten Vorwürfe nicht

Das nun vom Portal »The Decoder« veröffentlichte Urteil wurde vom Landgericht München I bereits am 28. Mai verkündet. Die Google vertretenden Anwälte argumentierten vergeblich, dass der Anbieter nur als »mittelbarer Störer« hafte und demzufolge nur dann tätig werden müsse, wenn offenkundige Rechtsverletzungen vorlägen.

Die Richter führten aus, dass es sich bei der KI-Zusammenfassung nicht nur um einen klassischen Suchmaschinenschnipsel handele, der lediglich eine Vorschau auf eine verlinkte Website darstelle. Stattdessen sei der KI-Text ein eigenständiger Beitrag Googles, für den das Unternehmen unmittelbar einzustehen habe. Dafür sprach auch, dass die von Google verlinkten Quellen einige Vorwürfe aus der vermeintlichen Zusammenfassung gar nicht stützten, sondern sich teils auf ganz andere Unternehmen bezogen.

Das Gericht ließ auch das Argument nicht gelten, dass Nutzende wissen müssten, dass man KI-Zusammenfassungen überprüfen müsse. »Im Übrigen müsste dies gerade auch den von der Verfügungsbeklagten vorgetragenen Nutzen der Funktion deutlich schmälern, wenn die ›Übersicht mit KI‹ allgemein anerkannt als nicht belastbar zu behandeln wäre und jeweils sämtliche der angezeigten Verlinkungen doch eigenständig zu prüfen wären«, heißt es im Urteil. Lediglich in zwei Punkten wurde die Klage zurückgewiesen, Google muss 80 Prozent der Gerichtskosten tragen.

Quelle: https://www.spiegel.de/netzwelt/kue...haften-a-511f1fb6-0f07-4de1-abc3-b7c9e3b8268f

 

Well, I never claimed otherwise, right? My post is literally a list of questions about this specific case and the distinctions this specific court drew. That said, rulings from any court (regardless of instance) can and will be cited as references by plaintiffs in future cases, which courts tend to welcome as it simplifies their decisionmaking. So while this is technically a case-by-case ruling, its implications don't necessarily stay that way.

This reminds me a bit of the time when labels and collecting societies hired law firms to flood P2P networks with songs and albums, then sent cease-and-desist letters to everyone who had downloaded and redistributed one of those songs or albums to collect 100 bucks per song or album. Great distribution strategy with a nice revenue boost.

PS: Ich lese grundsätzlich keinen Spiegel!

 

Since AI is relatively new territory and Google's AI operates and is accessible worldwide, I don't think the German ruling will become a major reference point; Germany is too small for that. The situation is different with the EU, though; it even has its own legislation for AI: EU member states have passed the world's first law regulating AI. To implement this into national law, the German government is creating an innovation-friendly legal framework effective February 11, 2026.

How should the court have ruled, then?
Everyone has the right to defend themselves against statements written to their detriment.

It's great that you remember how things were back then!
But I can't draw a parallel between this case and the wave of P2P-related legal warnings!

Today's business model for generating profit involves insults directed at politicians—something worth thinking about.
Habeck is filing over 700 criminal complaints regarding hate speech.

 

I think you've missed the point of what I was saying.
The issue isn't how a german court's first-instance ruling is received internationally, but rather what implications the reasoning behind it has for future court rulings. Will people start trying to coax some kind of delusional statements out of support chatbots on the websites of small or large companies in order to make a quick buck in court?

And here I disagree with you: since it is one of the first rulings of its kind, it will very likely have some international significance in court proceedings.

I haven't laughed that hard in a long time!

 

Hello @Sinus Well, first of all, I really want to praise your intelligence and foresight—I’m truly impressed! I don’t want to argue or dispute what you’re saying, but I’d like to add the following: "No plaintiff, no defendant"—someone has to file a lawsuit first, and then the court decides whether or not to admit the case. Whoever loses the lawsuit pays the legal costs, and ordinary people usually don’t take on the giants because such a trial can simply be dragged out until the opposing side goes broke. The lawsuit against Motorola started with a $150 million claim in the USA.

You're right about that—it's a real riot. The internet is a very difficult thing to regulate and is rife with cybercrime; they’ll probably turn to AI or else introduce that dreaded total surveillance of citizens—who knows what’s coming?