this post was submitted on 29 Oct 2025
23 points (100.0% liked)

Technology

1302 readers
97 users here now

A tech news sub for communists

founded 3 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
[–] CriticalResist8@lemmygrad.ml 7 points 4 days ago (1 children)

(posting my own comment so it's separate from the post body)

In the past most of these lawsuits ruled in favor of the AI company, when based on the grounds that content was used for training data. The common cited reason by the courts is they consider it fair use, though there can be other laws intervening depending on how they acquired the books. This was the reason Anthropic ended up buying their books and destroying them afterwards, just to make sure they stayed under fair use, while Google relied on academic sources and public libraries.

There's also a very important element to remember about such lawsuits, two in fact.

  1. even if openAI gets sued into oblivion, it will lead to their IP being swept up by other companies (see the other recent article about microsoft acquiring a 27% share in openai). The products such as Sora, the model files and the H100 GPUs won't disappear, they'll be sold to pay for liquidation if that happens.

2... Martin is a big author, and clearly he wants to protect his IP, which he owns, and he uses copyright law for it - the same copyright law that Disney, also suing an AI company (Midjourney), spent decades lobbying for. You can thank Disney for the idea that copyright extends 75 years after its creator's death so they could keep Mickey Mouse.

The same law he cites here against openAI can also be used against fan art, fanfics, and anything that he feels looks too close to Game of Thrones. It won't be a win for small artists against AI if he wins (which he is unlikely to, as seen from previous lawsuits), but a win for famous (and rich) artists who can afford to protect copyright. Small artists can't afford to go up against openai or anthropic, for any reason, but budding artists who may use AI in their process (including open source chinese models - but Martin won't come after deepseek but after the person who used deepseek to generate a name like Karlhisee).

More broadly, outside of lawsuits there are attempts by IP holders to reign in AI by 'teaming up' with small artists, whose labor is already exploited by these big companies who freely steal their creations to make money on them (see for example Zara stealing designs online to put them on their t-shirts). Copyright is a fuck.

[–] ComradeSalad@lemmygrad.ml 3 points 4 days ago* (last edited 4 days ago) (1 children)

The same law he cites here against openAI can also be used against fan art, fanfics, and anything that he feels looks too close to Game of Thrones.

I don’t see how this point is entirely relevant seeing as how such works are explicitly protected by non-commercial fair use rulings. Rulings surprisingly maintained by explicitly capitalist courts.

In the case of the United States, courts also hold a generally unfavorably view to artists or corporations claiming copyright infringement simply because a work “strongly resembles” another that is copyrighted. Making a case that a work resembles another is nearly impossible since the counterclaim is usually that the work is merely using common tropes, themes, or conventions. Even if an artist intended to bog down a smaller artist in endless litigation, the chances of a lawsuit like that making it past preliminary arbitration seems minimal at best seeing as how the case law is a settled fact, and I’ve yet to find anything courts love less than someone endlessly filing the same frivolous suits.

[–] CriticalResist8@lemmygrad.ml 5 points 4 days ago

Yeah, it was an unfinished point lol I agree.

In the US where Martin is filing the argument he makes that models are being trained on his works has already been litigated - the courts consider training to be fair use. Whether this is good or bad is something else but he's probably going to lose if he tries this angle, because it's been tried before and now there's precedent. However the courts have sometimes sided with the plaintiffs on other things, for example how the work was obtained (and this is why anthropic bought scanned and then destroyed books, so that they owned their copy and made private copies, legally standing. Legally speaking this was perfectly okay to do.)

What big media companies that live on IP, like Disney or hollywood studios, are doing is to try and restrict what counts as fair use so that they can stop AI companies from training on their IP. But if they manage to do this, they will also effectively be able to stop anyone else they already want to stop but can't - transformative art, remixes, non-commercial usage, and lots of other things I'm probably not thinking of.

The trap is they lobby for this by enlisting the free labor and money of 'human artists' (their wording) to pressure the government into making harsher IP laws under the guise of stopping AI training. But small artists won't be able to defend their creations even with harsher laws, because it's not cheap to defend copyright, and they don't have the means to scour the web for infringements anyway. What this will do, conversely, is reinforce companies like Disney and prevent these same small artists from making anything that these companies are not in favor of. Nintendo for example is a famously hated litigator over IP.

I quote a lot from the Artisanal Intelligence essay but they have a section on this phenomenon:

That essay perfectly encapsulates what Martin is doing with this lawsuit 2 years before the fact:

It's not random chance that Martin is the one suing - he owns the rights to A Song of Ice and Fire, not his publisher, so he's the one who would be owed money for IP infringement .