.png)
AI Can Write the Words. But Can It Really Be the Author?
- Rob Frankson
- 1 day ago
- 4 min read
AI is forcing publishing to answer a question older than the technology itself: what actually makes someone an author? A new academic argument published this month says the answer is not simply who, or what produced the sentences. It is who can take responsibility for them.
Quick Shortcuts
AI can generate words. Can it be an author?
That distinction matters because publishing has spent much of 2026 wrestling with AI-generated manuscripts, disclosure rules, copyright uncertainty and unreliable attempts to identify machine-written prose. The debate often starts with the text itself: did a human type these words, or did a model generate them? But that may be the wrong place to draw the line.
A research paper published on 12 August in Philosophy & Technology argues that treating AI itself as an author is a category mistake. Its central point is useful well beyond academia: authorship involves accountability. A statistical system can generate language, but it cannot meaningfully stand behind a claim, defend a creative decision, accept criticism or take responsibility for the consequences of what it publishes.
That sounds philosophical until you look at what publishers are actually doing. Oxford University Press says generative AI does not qualify as an author and should not undertake primary authorial responsibilities such as generating arguments, analysis or conclusions. It also places accountability for accuracy, integrity and originality firmly with the human author.
Publishing is moving towards responsibility, not detector scores
This is important because the industry still does not have a dependable technical test for authorship. HarperCollins chief executive Brian Murray recently described the situation as very murky and argued that publishers, agents and authors may need shared best practices rather than relying on AI-detection software. He even raised the possibility that writers may need to retain records of their creative process.
That approach makes more sense to me than treating every encounter with artificial intelligence as equivalent. A writer asking a machine to produce a novel and then presenting that output as entirely their own work raises very different questions from an author using AI to organise research, explore a visual concept, prepare advertising, improve metadata or handle some of the administrative workload surrounding a creative career.
The line I care about is creative responsibility
I write my fiction. The characters, worlds, narrative choices and finished stories are mine. At the same time, I am perfectly comfortable using AI around the business of being an author for research support, visualisation, advertising, online content, organisation and other tasks that help an independent writer actually reach readers.
For me, the useful boundary is not whether AI touched the workflow. Modern writing already passes through software full of automated systems: spelling correction, grammar suggestions, search, recommendation engines and increasingly generative features. The more meaningful questions are: Who conceived the work? Who made the substantive creative choices? Who reviewed the finished result? And who is prepared to put their name on it and be answerable for it?
That last question is especially important. Authorship is not merely a claim to credit. It is an acceptance of responsibility.
Human authorship could become more valuable, not less
There is an interesting consequence here for working writers. As synthetic text becomes cheap and abundant, readers may place greater value on knowing that a human being actually imagined, struggled with, revised and completed the book they are reading.
That does not require pretending technology is absent. In fact, transparency may become part of the value proposition. An author should be able to say clearly: this is the work I created; these are the tools I used around it; and I remain responsible for the finished result.
Creative provenance, notes, drafts, version histories, research trails and editorial decisions could therefore become less like defensive paperwork and more like part of a book’s story. It shows how an idea became a manuscript and where human judgement entered the process.
The real AI authorship test
Perhaps publishing has been asking the wrong question. Instead of endlessly trying to determine whether a sentence looks as though AI might have produced it, we could ask something more useful: is there an identifiable human creator who exercised meaningful creative judgement and accepts responsibility for the finished work?
AI can be a powerful tool inside that process. It can also be misused to replace it. Those two things should not be collapsed into the same category.
The publishing industry now needs standards sophisticated enough to recognise the difference.
Today’s question
If authorship ultimately means creative responsibility rather than simply producing words, where would you draw the line between an author using AI as a tool and AI effectively becoming the writer?
Keep up with AI & The Author
About Rob Frankson
Rob Frankson is a science-fiction author and creator of the Near Galaxy Saga. Through 121 Minutes, he writes about storytelling, publishing, technology and the changing creative landscape.
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