.png)
When AI Money Reaches the Author: Who Actually Owns the Rights?
There is a difference between using artificial intelligence as an author’s tool and using it as an author. Publishing is beginning to discover that the difference is not philosophical. It affects trust, discoverability and the basic economics of selling books.
This week that argument became harder to ignore. The Authors Guild renewed pressure on publishers over the $1.5 billion Anthropic copyright settlement, asking them to relinquish claims on long-out-of-print titles where rights have reverted and to compensate authors where publishers failed to register works. The dispute is a reminder that AI is no longer sitting outside publishing. It is now colliding with contracts, rights records and money.
In this article
Why the Anthropic settlement matters to working authors; why rights records suddenly matter again; the difference between AI-assisted and AI-authored work; what authors should do now; and why transparency has to work in both directions.
When AI money reaches the publishing contract
The Anthropic settlement was approved in July after authors accused the company of misusing books while developing Claude. The legal story is complicated: the case drew distinctions between model training and the acquisition of books from pirate sources. But the practical consequence is simpler. A very large pool of money now has to be allocated across copyrighted titles, and that has exposed an old publishing problem in a very modern setting: who actually controls which rights?
For authors, that matters beyond this particular case. AI companies and publishers are already exploring licensing markets. The UK Publishers Association reported earlier this year that AI licensing is established and growing, including licensing for text and data mining, training and retrieval-augmented generation. If books are becoming valuable inputs to AI systems, then accurate ownership records and understandable contracts stop being administrative trivia. They become part of an author’s economic position.
The distinction I keep coming back to
I use AI. I make no secret of that. I use it to help research subjects, organise material, visualise ideas, develop marketing and deal with some of the practical work that surrounds being an independent author. I also use it to challenge my thinking. None of that means I want a machine to write my novels for me.
That distinction is becoming more important, not less. An author using a tool to check research, explore a visual concept or organise a campaign is still responsible for the creative decisions. Handing a prompt to a system and publishing the resulting prose as a novel is a different proposition. The first can support authorship. The second can substitute for it.
Even formal publishing guidance increasingly recognises this distinction. UK government copyright material describes AI-assisted works as works where the creative essence comes from a human creator, while Oxford University Press tells authors that generative AI does not qualify as an author and stresses human accountability for accuracy, integrity and originality.
Trust may become the scarce commodity
The publishing industry has spent much of the last two years asking whether AI can write. I increasingly think that is the less interesting question. Of course it can produce text. The harder question is whether readers, publishers and authors can establish enough trust to know what they are buying, licensing or rewarding.
That trust problem cuts both ways. Writers should be candid about material AI use. If a machine generated the prose, readers deserve to know. But authors should also be able to discover when their work is being licensed, who controls that licence, what rights are being asserted and how any resulting payment is divided.
The Anthropic settlement is useful precisely because it turns an abstract argument into a ledger. Once there is money attached to the use of books, vague assumptions about rights become much harder to ignore.
What I would do as an author
First, keep your contracts and rights reversions organised. If an older title has reverted to you, make sure you can prove it. Second, understand the terms of any AI tool before putting unpublished material into it. Third, keep your own boundary between assistance and authorship clear enough that you could explain it to a reader without embarrassment.
And finally, do not assume that being interested in AI means surrendering your rights. I think authors can use these tools intelligently while still insisting on consent, attribution, transparency and fair treatment when our books create value for somebody else’s system.
The bigger question
AI is not going back into the box, and I do not particularly want it to. Used properly, it can be an extraordinary creative-business tool. But the publishing industry now needs rules that recognise two things at the same time: useful AI assistance is legitimate, and human creative work still has owners.
The Anthropic settlement will not settle the wider AI copyright argument. What it may do is force authors and publishers to look much more closely at the paperwork beneath it. And that may prove unexpectedly important as licensing becomes a real market rather than a hypothetical one.
What do you think: if your book creates value for an AI system, who should control the deal and how should the money be divided?
About Rob Frankson
AI & Editorial Transparency
AI & The Author is edited and published by Rob Frankson. Artificial intelligence is used to assist with news research, initial drafting, content organisation and supporting imagery. All articles are reviewed and, where necessary, edited by Rob Frankson before publication. The opinions, editorial position and final decision to publish remain the author's.
Comments