What’s the latest on AI & Author Copyright?
The US cases of Kadrey, and Anthropic
If you are a subscriber… welcome to the first emailed version of this new Substack! Thank you for joining me.
This post is a bit techy but I’m aiming to have a range of content; from industry knowledge to advice (via a few whinges). Just read the posts that work for you.
Before I start…
At the Harper Collins party last night Charlie Redmayne was talking about AI and how we must have fair copyright and payments systems in place (YES) but he was also saying AI is a tool and we should learn to make use of it.
I’ve had a lot of chats with people about the ethics of an individual person using AI. We all know it’s bad for the environment, bad for writers/illustrators/translators etc. So does that mean that, for moral reasons, we shouldn’t use AI at all?
Personally, I wouldn’t say that. For me, the onus should be on the corporations (who’ll make millions from the tech) to find climate-friendly, copyright-compliant ways to train and run their machines.
But I CAN see the power of NOT using it, and of forcing change through personal action.
What do you think?
Anyway, here we go ... The latest on AI:
What’s the Background
If you missed the background to this, essentially …
A pirated site (LibGen) was already illegally hosting millions of books. My work was on there. Your work was probably on there. These pirate sites are really hard to take down. The Society of Authors and others have tried many times over the years but, even if they succeed, the sites come back up again like Whack-a-Moles.
AI companies including Meta (the highly valuable company that owns Facebook, Insta and others) and Anthropic took over 7 million books on LibGen and used them to train their AI machines called LLMs (large language models). LLMs use the work created by authors to analyse patterns of text ie which word often follows another word.
What’s worse is that email evidence emerged showing that both companies particularly wanted to use BOOKS to train the LLMs because our work is of much higher quality than general internet sources (one Meta employee said ‘the best resources we can think of are definitely books’) BUT they decided it would be too costly to pay us (massive eye-roll), so they deliberately took pirated books.
I’ve got an article in The Independent covering a lot more here (paywalled).
https://www.independent.co.uk/voices/author-protest-meta-books-training-ai-b2726853.html
In June, The Data (Use and Access) Bill passed in parliament – unfortunately without the amendments fought for by Baroness Kidron (which would have forced AI companies to disclose what material they were using to develop their programmes). BUT Peter Kyle (Labour MP and Secretary of State for Science, Innovation and Technology) promises to add safeguards into the new legislation he has planned. We’ll see.
Can we sue in the UK?
Lawyers seem to agree that, under UK law, the META acts would amount to a breach of copyright… but it’s not that simple to sue from here. It depends on where the actual breach took place, where your book is registered and other complex issues. That’s why most of the cases are currently happening in the US, rather than the UK.
If the AI tech firms were based in the UK, it MIGHT be easier but at the moment, they are not. Peter Kyle is trying to get them to move here… more on that another time.
Aside from all those issues, it’s incredibly slow and incredibly expensive to sue.
OK, what about the US?
I’m not trained in US law but, roughly …
The US has a different Copyright Act, which has an alternative version of ‘fair dealing’ (the UK term). In US law, it’s called ‘fair use’, and there exists a defence that we don’t have in UK law. The defence allows limited use of copyrighted work, weighing up 4 factors. One of these is whether the new work is ‘transformative’, another is ‘the effect on the market for, or value of, the work’ ie whether the use would cause market harm.
Get to the point. What has happened in the caselaw?
Judgments were given in two US cases last week: Bartz v. Anthropic and Kadrey v Meta.
Anthropic:
This case involved the taking of authors’ work to train the Claude AI LLM. Anthropic used both print texts AND texts from pirate websites to train Claude.
The U.S. District Court for the Northern District of California held:
1. Training Claude on legally acquired print books was fair use – therefore no copyright breach.
2. Downloading and copying pirated books was not fair use – and the case will proceed on this issue.
Obviously, point 1 is hugely disappointing – but US lawyers expect it will be appealed because it seems to go against established US case law. Watch this space.
Further, the judge made the following ridiculous argument:
‘Everyone reads texts, too, then writes new texts. To make anyone pay specifically for the use of a book each time they read it, each time they recall it from memory, each time they later draw upon it when writing new things in new ways would be unthinkable.’
I’ve heard this nonsense before and responded to it via letter in the Guardian here:
https://www.theguardian.com/technology/2025/apr/09/ai-doesnt-care-about-authors-but-meta-should
Kadrey v Meta
This case was about Meta taking pirated books from LibGen to train the Llama AI LLM.
First the good news. The judge (in a federal court in California) said
‘…by training generative AI models with copyrighted works, companies are creating something that often will dramatically undermine the market for those works, and thus dramatically undermine the incentive for human beings to create things the old-fashioned way.’
But the court held that the plaintiffs failed to provide sufficient evidence of market harm (as required under the US Copyright Act – see above). Empirical or expert evidence was needed on this point.
So TECHNICALLY it was a win for Meta – but not really because the courts are hopefully beginning to recognise that this DOES harm us, which will help in future cases.
I’ll try to keep you updated on it all but it’s very fast moving.
Thanks everyone.
Next time, I going to write about
‘Help! Someone has the same book as my work in progress’
Happy writing!
A x