.png)
When Publishing Draws the AI Line: A Useful Rule for Authors
There is a difference between using artificial intelligence to help an author work and asking artificial intelligence to be the author. Publishing has spent several years struggling to describe that difference. This week, South Korea’s publishing industry has made a serious attempt to draw the line.
The Korea Publisher Society has unveiled a set of principles for publishers and authors dealing with AI, developed with writers and critics as concern grows over mass-produced “click-to-publish” books. The important part is not that another organisation has produced an AI policy. It is the distinction underneath it: human creative responsibility remains central, while AI can still be used as a supporting tool.
In this article
Why publishing is beginning to distinguish AI-generated books from AI-assisted work
Why human responsibility matters more than a blanket ban on technology
How authors can use AI without handing over authorship
What this shift could mean for readers, publishers and creative trust
The problem is not the tool. It is who is doing the creating.
I have never found the argument that authors must either reject AI completely or surrender the creative process particularly useful. It confuses two very different things. If I ask an AI system to generate a novel and put my name on the result, I have handed over a substantial part of authorship. If I use AI to help research a technical idea, organise notes, test a marketing angle, visualise a fictional location or deal with the business surrounding my books, I am still making the creative decisions.
That distinction matters because the technology is already embedded in publishing. Publishers Weekly recently reported that publishers are using AI in back-office work including metadata, proofreading, royalty administration and customer service. Those uses deserve scrutiny, but they are not the same as flooding a bookshop with machine-generated novels.
Readers are buying a promise as well as a book
When somebody buys a novel, there is an unspoken transaction taking place. They are not simply buying eighty thousand words arranged in a plausible order. They are buying another person’s imagination: choices, obsessions, mistakes, humour, experience and voice. That is particularly true of fiction. A story matters because somebody meant it.
This is why mass-produced AI fiction creates a different problem from an author using an AI assistant. The issue is not merely quality. Cheap automated production can overwhelm discovery systems, blur provenance and make it harder for readers to know what they are buying. A publishing market that loses confidence in authorship eventually damages genuine authors along with everybody else.
Copyright is making the distinction harder to ignore
The legal argument is moving as well. This week a US appeals court upheld Thomson Reuters’ copyright victory against Ross Intelligence over the use of Westlaw headnotes to train a competing AI legal research system. It is a different market and a different set of facts from fiction publishing, so authors should be careful about treating it as a universal answer. But it is another sign that courts, businesses and creators are being forced to define where legitimate technological use ends and the exploitation of protected work begins.
My line is still a simple one
I write my fiction. The worlds, characters, narrative choices and final prose are mine. I do not want a machine to replace the part of writing that makes the work mine in the first place. But I am equally unwilling to pretend that useful technology becomes illegitimate simply because it contains AI.
Research support can save time, provided the facts are checked against trustworthy sources. Visualisation can help me explore a scene or communicate a fictional world. AI can help organise a campaign, analyse options, prepare material for different platforms or take some of the repetitive work out of running an independent author business. None of that requires me to outsource the novel.
The Authors Guild has taken a similarly practical approach in its model contract clauses, addressing AI training, subsidiary rights, translation, narration and an author’s own use of AI rather than pretending one rule can cover every application.
A rule readers can understand
That is why the South Korean principles are interesting beyond South Korea. Publishing needs rules that ordinary readers can understand. “No AI” sounds simple but becomes increasingly meaningless when AI features sit inside software used for editing, search, design, analytics and administration. “Anything goes” is worse.
A better starting point is human authorship, disclosure where AI materially creates content, respect for copyright, and human accountability for what reaches the reader. That gives authors room to use technology without allowing technology companies or automated publishers to redefine authorship out of existence.
The interesting question is no longer whether AI belongs somewhere in publishing. It already does. The question is where we draw the line — and whether readers can still tell when there is a human being on the other side of the page.
Where would you draw it? Which uses of AI feel like legitimate author support to you, and which cross the line into replacing the author?
About Rob Frankson
Rob Frankson is a science-fiction author and creator of the Near Galaxy Saga and HMSS Broadside. At 121 Minutes he writes about storytelling, publishing, emerging technology and the practical realities of using AI around a human-led creative process.
Keep reading AI & The Author
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