I literally laughed out loud reading the headline. Good shit, hopefully the Find Out season will carry on at this kinda pace. Probably won’t, but it’d be nice to see.
A German regional court has ruled that Google is directly liable for false claims in its AI-generated search overviews.
Unfortunately, the regional court is the lowest court stage. This will climb up until highest German court and after this to the EU court, I expect to see.
Being then a „Grundsatzurteil“ that is leading all courts in Germany. Our legal system isn’t case driven.
Unfortunately, the regional court is the lowest court stage
No. Landgericht is the second stage already.
It’s Germany, they’ll just find a way to blame Brussels and throw more money at the US as an apology.
A regular search engine just points to outside websites. But AI overviews generate “independent, new, and substantive statements” by evaluating and combining content from various third-party sites. And only Google can check those statements, the court said, “at least by comparing the underlying third-party websites with its own statements based on them.”
Honestly this is all the reasoning you need to infer that Google should be liable. Google alone has editorial control over the summary their AI generates, not the outside sources used to generate these statements, ergo Google should be held liable for that.
At the hearing, Google argued that users could check the linked sources themselves to verify whether the AI summary was correct. Users generally knew “that information generated with AI should not be blindly trusted,” the company claimed.
… And you know that’s true when the best Google could muster as a defence is to say that people shouldn’t be blindly trusting the AI, which ironically means even Google thinks their AI is full of shit.
But unfortunately for Google, not only does the court not buy that defence, but it would appear that’s contrary to how most people use the feature.
The ruling may also have international reach, according to the court.
I seriously hope so. Its about time companies started taking proper liability for the actions of their LLMs.
And if Google complains that it’s on the pieces of info they got from 3th parties that were wrong and name them, then the 3th parties are able to request compensation for using that info.
Exactly. Can’t have it both ways.
If Google want to claim the liability falls with the source’s its pulling from, then it should be taking explicit permission to cite these sources and be paying them.
Otherwise it’s an AI-powered editorial, and that’s on Google.
Though personally I’d be happy with the entire system being scrapped, as it only serves to fuck over small publishers and people’s ability to search for and be critical of information.
deleted by creator
I fail to understand why it should be bad for small companies.
In my experience most small companies don’t have public AI summaries. And even if they do i still think it’s their obligation to check what they make public.
In the not so distant future just about every site will have AI summarization or QnA as a core part.
Instead of searching through endless documentation you ask AI to trawl and give you the answer. This is undeniably useful. But if they give the wrong answer once and suddenly become liable, that’s a potential risk.
How is it “undeniably useful” if it has the potential of giving wrong answers?
Also and perhaps more importantly, are these the lengths people go to avoid reading? If so, we are doomed.
Not everyone enjoys reading documentation. We don’t need to be defensive about this. We already have search that can trawl through a well maintained site.
AI can not only go through the documentation but also translate it to layman and point to the sources.
If it gives the wrong answer 1 in every thousand results, it is still undeniably useful. You shouldn’t blindly trust AI is common place knowledge. And it’s no different than doing a Google search for something and some times clicking into a result that is bad. The fact that that possibility exists doesn’t change the fact google is "undeniably " useful.
I’m going to be straightforward with you and say that if someone doesn’t want to read documentation, they shouldn’t be doing the job the documentation is for.
I’ve been bitten by AI summarizing documentation so many times, these days I refuse to use it for that purpose anymore. It’s just not worth it. It creates a loop where it wants to try things that don’t work, walk back, try something else, repeat, and spend $10 worth of tokens in the process.
You say that I shouldn’t blindly trust AI like I shouldn’t blindly trust Google results. The difference is that AI is presented as an authoritative source in itself. Hell, most of the time LLMs don’t link sources unless explicitly asked for. And here’s the thing, if I have to go and read the actual sources, it isn’t doing anything significantly more time efficient than just text search, but it is doing it at ten times the cost.
In the not so distant future just about every site will have AI summarization or QnA as a core part.
Hopefully not, and this ruling goes some way to ensuring sense prevails. It’s a little different if the LLM providing the “AI” summarization has been trained exclusively on the contents of the site; that ensures that only the work of the site authors is used in generating the summary, which means it’s their words, and also probably less likely to hallucinate.
Instead of searching through endless documentation you ask AI to trawl and give you the answer. This is undeniably useful.
I deny it. The results of an LLM being used to answer a question are far too often wrong to ever be trusted. Sometimes the errors are obvious, much more often they are subtle and harder to spot, but delivered with certainty none-the-less. This ruling ensures that the ones providing the LLM summary are held liable, in the same way they would be if a human wrote the same summary.
But if they give the wrong answer once and suddenly become liable, that’s a potential risk.
Correct, and that is as it should be. Apply the same logic to a human written piece and you will see that.
A potential risk that any company implementing an AI for something as simple as a Q&A should be aware of prior to doing that.
If they don’t want the liability, then just don’t use AI for public facing functions. Its not difficult.
In the not so distant future you use your own personal AI to do the trawling and if that thing gets it wrong that’s on you or the company that made it.
I have a feeling that the megacorporation’s AI generating false statements about smaller businesses that effectively drives customership away from them harms a lot more small businesses way more than AI-powered businesses being held to account for what their AI states as fact publically.
(And that’s not even counting the harm Google’s AI summaries are already doing to small publishers by driving traffic away from the very websites its using as source material.)
If a company doesn’t want the liability associated with a rogue agent making false statements, then I’ve got news for you - they don’t have to use AI. Literally nobody is forcing small private businesses to use AI for anything.
And what @MyButSmellsBat@feddit.org has said is entirely true. Most small companies won’t have an AI, and those that do should still be held accountable for their AI’s public statements.
the AI makes its own claims that don’t appear in any linked source, and the operator has to answer for them. […] if it gains traction internationally, the fallout could hit not just Google but every AI provider
And that is a good thing!
We (the world) need at least some basic level of quality and truth in AI generated answers. FINALLY.
This appears to be impossible with current LLMs. You would need an actual human to verify every possible search result as the LLM is incapable of doing that for itself
appears to be impossible with current LLMs
Not the court’s problem.
“Sorry, your honor, my weapon is that faulty so I can never know who it is who will be killed, but I just had to shoot because that’s how I make my money…”
That’s my point, the problem is the LLM itself shouldn’t even be being used to begin with. I’m not defending AI bullshit by any means. I’m saying “truth” or “quality” are not qualities that an LLM will ever possess by its own nature. The ultimate solution for truth or quality is no LLMs, but I guess that ship has sailed.
Sometimes it does not even matter if it is truth or not.
That may actually be such a case here: Factual statements that can create bad reputation for somebody (or some company).
deleted by creator
This isn’t final. Google has time to appeal. Let’s hold off on the label “landmark” until it reaches legal effectiveness. Which it probably won’t, however good a verdict by a German regional court, much less one based in Bavaria, this is in my opinion.
Google lawyers arguing in court that Google’s so-called AI results are shit anyways and people should know it is chef’s kiss.
I don’t know where you’re from, but typically within the EU, especially in countries like Germany, Google and other mega corporations from the US don’t have that much sway (yet) within the justice system. I wouldn’t be surprised if this is validated in the near future by more impactful courts.
I think my sniping at Bavaria speaks for itself.
They don’t need sway as much as money and lawyers, which I imagine they have. And this verdict is probably on the worst outcome end of the scale for them. I cannot imagine they will accept a ruling that calls them daft like this one does. They will try to water down liability for their model’s fantasy summaries. Whether they succeed is a different question. But they will try, so they will appeal, so this verdict isn’t worth the paper it’s printed on. Yet.
All I said is that this verdict isn’t effective yet. These headlines and sadly this article buries this fact in a sentence in the last paragraph. Blink and you miss it stuff. Lemmies tend to overlook this and declare victory over Google when this was merely the first battle of the war.
It will be a “landmark” case, regardless of what the outcome is.
I mean yeah? This is the only way to square the circle. Same as if you buy a thing from amazon, it does not matter what they try to pull in the back you went and bought a thing from a place. If you google a thing and google shows a wrong (and often plain dangerous) answer then yeah, that counts as google! Maybe if they did not also try and fake the result being true they could have an argument.
And there is already precedence for this as a few nation’s courts have found that a company is bound by promises made by their own AI agents that it empowers to answer customers. This is just the same idea but for search. I hope it goes though all the German courts and is picked up in other places.
Interesting showstopper for the AI-Bubble. Let’s see where this is going.
I don’t use Google, but DuckDuckGo Search Assist often shows summaries with links that don’t at all support what the summary says, but I often still forget to fact check it. I might start using the no AI version of the site because I’m concerned I might accidentally register something important as factual at some point that isn’t.
All they need to do is use the exact text from the answers and attribute them to the source…
