top of page
Search

AI Is Testing the Author’s Identity: Copyright, Detection and the Fight for Human Work

Writer: Rob Frankson
Rob Frankson
45 minutes ago
5 min read

AI & THE AUTHOR — WEEKLY ROUNDUP | 14–20 SEPTEMBER 2026


This was the week the argument about artificial intelligence and writing became much more personal. The question is no longer simply whether AI belongs somewhere in the publishing process. It is increasingly about whether readers, publishers and platforms can still tell what is human, who owns the material feeding the machines, and what happens when technology starts borrowing an author’s identity as well as their words.


The week in one sentence


AI is becoming ordinary infrastructure around publishing at the same time as its use in the creative work itself is becoming more contentious — and that distinction matters enormously to working authors.


1. The copyright argument acquired uncomfortable new evidence


One of the most consequential developments came from newly reported court material in the continuing copyright battles around AI training. Reporting this week described internal discussions concerning the use of pirated-book sources and copyrighted publishing material during model development. Whatever the courts ultimately decide about fair use, the disclosures matter because they move part of the argument away from an abstract debate about how machine learning works and towards the practical decisions companies made when assembling training data.

The legal picture remains complicated rather than one-sided. On 16 September, a US appeals court upheld the dismissal of part of a software developers’ case against OpenAI and Microsoft concerning removal of copyright-management information, while allowing separate open-source licensing allegations to continue. That is a useful reminder for authors: there will probably not be one grand ruling that settles every AI copyright question. Different claims — infringement, piracy, licensing, attribution and contractual rights — can produce different outcomes.


2. AI detection is becoming a risk to human writers


For writers, perhaps the most disturbing story of the week was not about an AI-generated manuscript at all. The Independent reported the case of writer Jerry Falade, who says a major book deal collapsed after his use of African dialect was mistaken for AI-generated writing. The wider issue is bigger than one dispute: detection software is being asked to make judgements that can affect careers even though there is no universally accepted detector capable of proving that a text was written by AI.

That creates a strange inversion. Writers were initially worried that AI would imitate human writing. Now some human writers also have to worry that their own voice may be labelled artificial because it does not resemble the linguistic patterns a detector expects.

For authors, this makes retaining drafts, revision history, research notes and manuscript versions increasingly sensible. Not because writers should have to prove their humanity, but because provenance may become useful evidence in a publishing environment where automated accusations can carry real consequences. Read The Independent’s 17 September report.


3. The flood of machine-made books is becoming an economic problem


The commercial side of the argument also sharpened. Fortune reported on 14 September that AI-generated books are flooding online marketplaces, increasing pressure on discoverability and author income. The problem is not simply that poor books exist — publishing has never lacked those. It is the near-zero marginal cost of producing enormous quantities of synthetic material and pushing it into the same discovery systems used by human writers.

Then came another evolution of the same problem. The Atlantic reported on 18 September on so-called “parasite authors”: AI-enabled publishing operations that imitate established writers and attempt to capture readers searching for those authors or genres. That moves the threat from generic AI slop towards something closer to identity and audience appropriation.


4. Writers themselves are drawing the line at creative substitution


In the UK, screenwriter and Writers’ Guild of Great Britain president Jack Thorne argued this week that undisclosed use of AI to generate complete professional scripts amounts to cheating and called for stronger protections for writers. His intervention matters because it exposes the fault line that is likely to define the next stage of this debate: using AI around creative work is not necessarily the same thing as asking AI to replace the creative act.

That distinction closely reflects my own position as a working author. I use AI as a tool around the work — for research assistance, visualisation, organisation, marketing and experimentation. I do not regard that as equivalent to asking a machine to author my fiction. Microsoft Word already assists with spelling and grammar; digital tools have always changed how writers work. The meaningful question is not whether technology touched the process. It is who made the creative decisions and who actually wrote the book.


What changed this week?


Taken together, these stories show the debate maturing. A year or two ago, the loudest question was whether generative AI would replace writers. This week’s developments suggest a more complicated reality. AI is becoming useful infrastructure for administrative and supporting work, while the industry is simultaneously discovering that authorship, provenance, licensing and identity need stronger boundaries.

That is actually a healthier argument. “AI: yes or no?” was always too crude. A publishing workflow that uses automation to format metadata is fundamentally different from a synthetic novel masquerading as a human-written book. Research assistance is different from plagiarism. Generating a marketing concept is different from impersonating an established author. If the industry treats all of those things as identical, it will create bad rules — and potentially punish genuine writers while failing to stop industrial-scale abuse.


What authors should watch next


  • Copyright litigation: watch the difference between cases about lawful training, pirated source material, attribution and licensing. They are related, but they are not the same legal question.

  • Proof of provenance: keep manuscript drafts and revision histories. They are part of good creative practice anyway and may become increasingly valuable.

  • Platform discoverability: AI-generated volume and author impersonation could make reader trust, recognisable branding and direct relationships with audiences more important.

  • Disclosure rules: expect publishers, competitions and professional bodies to become more precise about the difference between AI assistance and AI-generated creative work.


The question for writers


If AI can help an author research, visualise and market a book without writing the book itself, where should publishing draw the line — and who should get to decide when that line has been crossed?


About the author

Rob Frankson is a science-fiction author and creator of the Near Galaxy Saga. Through 121 Minutes and AI & The Author, he explores writing, publishing, creativity and the practical impact of emerging technology on working authors.


Join 121 Minutes for new AI & The Author articles, publishing discussion and updates from the Near Galaxy.


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

Rated 0 out of 5 stars.
No ratings yet
Add a rating
bottom of page