If I post “heathrow83829 is a convicted poopoo head” (replace with your favourite crime) as if it were a factual claim, you’d be well within your rights to sue me for defamation even if people should apply some critical thinking skills and say “wait a minute, how would pdpi know that? Are you sure he isn’t just talking out of his arse?”
Now, search engines are usually afforded some amount of protection against defamation claims — they’re not held liable for simply indexing and quoting third party defamatory claims. Which is to say: Google wouldn’t be liable for claiming you’re a poopoo head if this comment shows up in search results.
The point of this ruling is that AI-generated text isn’t a quote from a third party, it’s text generated by Google’s own tools, so it’s speech by Google itself. It might be wrong, sure, but it’s still presented as a statement of fact.
At trial they can have the whole debate about whether Google was negligent in how they build their systems, and all that jazz, but let’s be clear here — it’s not a matter of every little factual mistake getting Google sued (and that would be absolutely terrifying from a freedom of speech perspective), but rather that the technical means by which you generate content doesn’t change your liability in publishing that content.
The other big point is that some people treat LLMs as if they are truth. We all know that they get a lot wrong, a lot of the time. Yesterday I watch both Elon post something somewhere between badly worded and wrong and another user used Grok to "prove him wrong", but this time being completely wrong in a different way. And this was a matter of law which can be looked up. Anything within throwing distance of politics won't be considered "correct" by everyone ever. Just defining "correct" is harder than it sounds and that probably isn't possible in reality.
We don't all know they get it wrong a lot. In fact, there's an overwhelming number of people who take it at face value
I see people claiming LLM are trustworthy on HN all the time.
It is not just some people, it is a lot of people insode the tech itself pushing the "trustworthy" claim.
I think Google added that AI-generated responses maybe incorrect? When you are paying such a low amount of cost, like probably for free, I don't think you can expect a same level of quality as a human written or reviewed of answer. It is like same random user spin up their Lovable and vibe-coded a piece of slop and hold Lovable responsible for not giving them production quality code. It is simple, you get what you paid for. If someone actually figured out AI that is actually always correct, it would be charged in superhuman price as well.
This case isn’t about Google’s users getting low-quality search results. As you say, you get what you pay for. The actual issue is that, in some searches, the AI summaries would claim the plaintiffs were guilty of scams and all-around shady business practices.
Put differently: it’s not newspaper readers complaining the paper is inaccurate, it’s the people mentioned in the articles.