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.
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.
To me it seems pretty reasonable that content Google generates with Gemini and publishes on their own google.com domain is their speech. Or at least they are fully editorially responsible.
When I wrote "Gemini is not illegal" it would have been more correct to say that the court has not decided about Gemini but specifically about search.
And I do think it makes sense to distinguish between explicitly chatting with an LLM and entering something in the search bar people have used for decades now.
I'm just wondering whether the logic of the ruling is tied to the specific context (search results) or applied more generally to LLM generated text.
The problem here is that
1) the AI Overview is giving incorrect information, and
2) it is Google's own words
...which makes Google liable for anything false there.
They weren't liable for false statements in search results, because that was 100% other people's content, being linked to, with a small blurb taken directly from their words. With the AI Overviews, regardless of whether it's getting its information from other people's content, it's remixing it based on Gemini's algorithms and synthesising statements that are frequently wrong.
If ai output is not copyrightable, it should not be considered personal output. So nobody should be responsible for it. Or if it is considered personal output, it should be copyrightable. Or perhaps the ai companies will be liable for all output, and they will therefore all cease to exist in any useful form? This seems like another alternative, where the output legal value is not central, but there will be a thousand different fights about how it is presented to others.
search engine -> search terms -> search command
Google accepted the argument up to that point. Then they argued that they are not responsible for the search result. The court argued that what they did was not a search (or not protected the way a search is).
The argument whether Gemini, ChatGPT or Claude are search engines and whether a prompt is search terms is a different argument. Certainly an interesting one, but not what that court looked at.
It's not just about the output, the input matters also.
