On this episode of Self-Publishing with ALLi, Dan Holloway covers a turbulent week for AI and publishing: the Association of American Literary Agents updating their Canon of Ethics to allow closer work with indie authors; a seven-figure publishing deal shelved within hours of an AI allegation; and a peer-reviewed study of 14,419 self-published Amazon titles finding that AI-generated books are commercially viable but outsold by human authors — while the number of titles grows twice as fast as revenue.
Listen to the Podcast: Literary Agents Update Ethics Rules
Sponsor
Self-Publishing News is proudly sponsored by PublishMe—helping indie authors succeed globally with expert translation, tailored marketing, and publishing support. From first draft to international launch, PublishMe ensures your book reaches readers everywhere. Visit publishme.me.
About the Host
Dan Holloway is a novelist, poet, and spoken word artist. He is the MC of the performance arts show The New Libertines, He competed at the National Poetry Slam final at the Royal Albert Hall. His latest collection, The Transparency of Sutures, is available on Kindle.
Thoughts or further questions on this post or any self-publishing issue?
If you’re an ALLi member, head over to the SelfPubConnect forum for support from our experienced community of indie authors, advisors, and our own ALLi team. Simply create an account (if you haven’t already) to request to join the forum and get going.
Non-members looking for more information can search our extensive archive of blog posts and podcast episodes packed with tips and advice at ALLi's Self-Publishing Advice Center.
Read the Transcript
Dan Holloway: Hello and welcome to another Self-Publishing News podcast. What an utter mess of a week — AI, indie authors, controversies, a very controversial viral study, armchair detectives. But before all of that, let's start with a slightly less controversial subject: the Association of American Literary Agents has changed their delightfully named Canon of Ethics to reflect, as they put it, the changing nature of writing — and in particular self-publishing and hybrid publishing — to enable agents to work more closely with indie authors in the way that indie authors would like them to.
Various things that agents are now permitted to do that they weren't before. The most prominent is providing or selling marketing services or non-editorial services — something that wasn't even featured in the Canon of Ethics previously. Now it is, and there are guidelines on how agents must be transparent when providing and recommending such services. Referrals must be clearly disclosed, and the guidelines specifically rule out using the slush pile as a funnel for selling marketing services rather than actually reading and considering submissions. Transparency is the watchword throughout.
One thing notably absent from the new guidelines: any provision on how agents and editors should be using AI when assessing or editing their clients' manuscripts. Given recent controversies — the Shy Girl case being the most prominent — this is one of the biggest outstanding questions in the industry, and it has been left unaddressed.
Seven-Figure Deal Shelved Six Hours After AI Allegation
AI has been at the center of monumental controversy this week, including two seven-figure publishing deals under the spotlight. One was shelved. The book in question was written by Jerry Folay and titled Call Me, I'll Hide the Body — it was subject to a massive bidding war. But within six hours of an allegation of AI use circulating, the publishers withdrew, issuing a statement saying they could no longer rely on what they had been told about how the book was created. The seven-figure deal for that title and a subsequent book were pulled.
This has caused a great deal of controversy and, needless to say, significant backlash from writers who feel — understandably — that this is beginning to look like a witch hunt. Armchair detectives are everywhere, claiming books were created by AI. We know that AI detection software is imperfect at best, and that many of the supposed tells are simply markers of good writing — ways people have always written, because AI is, after all, trained on how people actually write. And these are people's livelihoods that are now being affected.
A Peer-Reviewed Study of 14,000 Self-Published Amazon Titles — and What It Found
Which brings us to a hugely controversial peer-reviewed university study of 14,419 self-published titles on Amazon, from researcher Tuan Chakrabarti. It has created a major stir — in part because it identifies H.M. Wolfe, author of the highly successful Daggermouth, as the highest-profile title showing significant signs of AI involvement. Daggermouth, as I reported a month ago, is at the center of the emerging dystomance genre, and H.M. Wolfe has since signed a seven-figure deal with a Simon and Schuster imprint. Unlike in the Folay case, Wolfe's publishers have — at the time of recording — stood by their author and said they believe the book is human-authored.
The study is called Generative AI Floods and Dilutes the Market for Books. Its purpose is not primarily to finger individual authors but to test the hypothesis — put forward in various lawsuits — that the surge in AI-generated content is genuinely competing with human-authored content and reducing the financial share available to human writers. This is the hypothesis underlying the claim that of the four million books published in 2025, up from three million the year before, roughly one million were AI-generated.
Chakrabarti's key finding: across the span of the survey, the number of titles increased by around 20 percent, while the amount of sales increased by less than 10 percent. In other words, the rate of increase in the number of books is roughly double the rate of increase in revenue entering the system — meaning that on a crude per-title calculation, human-authored books are taking home less than they used to.
It's not quite that simple, though. Chakrabarti also found that sales were relatively lower for AI-generated titles than for human-authored ones. He estimates that around a quarter of the 14,419 titles he examined were AI-generated or significantly AI-assisted — a caveat he makes clear, acknowledging that AI detection software is imperfect and his findings are probabilistic, not definitive. If that 25 percent figure is correct, it aligns with the broader claim that roughly a million of last year's four million published titles were AI-generated. That doesn't mean it's right, but it is consistent with other figures in circulation.
Those AI-generated titles sold fewer copies than human-authored books — but not none. They are selling at commercially meaningful levels, not disappearing into the long tail of titles that sell fewer than ten copies. They are commercially viable, just not at the same level as human-authored titles. That gap is worth watching. If human authors continue to outsell AI-generated titles, it may be partly because of reader loyalty built up over years with known authors. Whether AI-generated titles could build similar loyalty over time — however strange that sounds right now — is a question we can't entirely rule out.
Thank you as always for listening, and I very much look forward to speaking to you again at the same time next week.
