As my wife keeps pointing out to me when I incessantly point out that “people on the internet are wrong about copyright”… who cares? The courts certainly don’t.

Courts have ruled against Google caching and displaying content in search timelines before. Google vs NewsCorp ultimately ended with Google paying NewsCorp license fees to index its content! But most people haven't sued because they actually want to be indexed by Google (and Googlebot respects robots.txt if they don't)

But Google search does not use images for the same purpose, and does not sell access to the images, and at present most AI models don't offer an opt-out, so the copyright implications are different (most logically, indexing and displaying an image with tacit permission and a backlink on an ad-supported web page is not the same as selling derivative works based on an image). By extension, they are not necessarily the same for StableDiffusion as they are for DALL-E, or for LLM models as they are for generative image models or licensed codebases, and copyright law has never been and probably never will be applied evenly across different jurisdictions, mediums and even individual cases.

I can't find any lawsuit between News Corp and Google. In what country? Google did agree to some kind of paid agreement with News Corp in 2001, according to news coverage at the time in part to avoid the threat of legislation in Australia that might have required Google to pay for content to index it. (No such legislation was passed in Australia, let alone the USA).

One USA case from back in the day that ruled that Google Indexing (and caching) was fair use, was Field v. Google, from 2006. http://www.eff.org/IP/blake_v_google/google_nevada_order.pdf

Another USA case that ruled that Google Image indexing, and using thumbnails in search results, was fair use was Perfect 10 v. Google (2007). https://cyber.harvard.edu/people/tfisher/IP/2007%20Perfect%2...

It's true it hasn't been super extensively litigated, and never made it to the supreme court. Perhaps because in fairly short order content owners realized they actually wanted to be indexed in google. And because Google (if not everyone) would respect a robots.txt to keep them out if they didn't.

The Google Books case is also probably relevant. (Which Google also, eventually, won).

There are other cases not specifically involving Google that are relevant, one for example with an index of tv news broadcasts that provided keyword search and let users view clips. The court ruled this was fair use too. Fox News v. TVEYES, Inc., 43 F. Supp. 3d 379 (S.D. N.Y. 2014).

I don't think I can find any USA cases that suggest indexing for a search engine would be a copyright violation and not fair use. Let alone against Google specifically. When you say courts have ruled against Google, can you cite any particular cases? Maybe they were from other jurisdictions than USA?

And yet, GP's description applies just as much to google indexing images. Google makes money off of a search index that copies copyrighted content without explicit permission or compensation.

In the USA, where OP suit is filed, I am not a lawyer but I think these image AI's are going to have a pretty good case for fair use, even if it what they did counts as "making a copy" (which I think isn't totally clear). It is a very transformative use. I think convincing a court that the AI's will result in lost revenue (for the specific works used as training data, not in general for the field) would probably be necessary, but not sufficient, to overcome a fair use defense -- and I don't think that argument is even a slam dunk. And is still just one factor in fair use; the transformative nature of the use has lately been given a lot of weight in the law, and seems pretty overwhelming here.

Of course, sometimes courts do surprising things.

News Corp fought Google mostly on their home territory of Australia, a battle which ended up with Google caving when News Corp succeeded in lobbying to get the law changed to be unambiguously in their in their favour (a battle AI providers should expect to face; the US isn't the only jurisdiction that's significant to them and legislation is bound to be updated for new tech). In France they got slapped with a fine of half a billion euros by an anti-trust court for not abiding by a French court order to agree fair license fees for news excerpts, and the EU's Copyright Article 13 is specifically designed to make platforms like Google News subject to EU-wide copyright law.

Google Images thumbnails were ruled infringing by one US court and overturned by another some time ago. More recently, the US also is the venue where Robin Thicke and Pharrell Williams ended up paying the Gaye estate royalties and naming Marvin Gaye a writer because Blurred Lines was apparently the result of Thicke saying he wanted to write a song that sounded like Got To Give It Up, and Williams using his knowledge of songcraft and what was distinctive about Got To Give It Up's groove to compose an original song with different words, note patterns and beat which nevertheless sounded similar, so I'm not sure there's much grounds for the absolute confidence most HNers seems to have that fair use defences will prevail in the US because of novelty and "learning".

Copyright law rulings are all over the place, and I think it's quite likely creators of such tools will be ruled infringing by one court and overruled by another, infringing in one region and OK in another, one diffusion model ruled non-infringing and another more commercially-oriented one ruled infringing, and perhaps others where either the creators or the users of the tool are deemed infringing but not necessarily both.

It doesn't matter what a lower court ruled once it's been overturned by a higher court.

As far as I know, all standing case law in the USA says that web crawling for search indexing is fair use. I'd be interested in a citation to a counter example. I've been unable to find any.

Also interested in specifics of any actual court cases involving News Corp vs Google in Australia; I can't find any references to lawsuits, only to the threat of legislation (which changes the law if passed, but I don't think it was passed in the end?)

The France thing is interesting. I know less about EU copyright law. Did Google pay the fine and... then what? Did they stop indexing news orgs in France, start paying for it, or just keep doing the infringing thing, or what? The ruling only applied to news organizations?