Honey, We Bought an AI Story
This post discusses suspected AI-generated fiction submissions, author verification, and their impact on small presses.
You can also read and download our full report: The Machines Are Coming for Your Masthead
Are you worried about AI?
We were asked this question on a small-publishers panel at the World Fantasy Convention in 2025. Our Managing Editor, C.L., remembers answering pretty confidently. AI is glorified predictive text. It can produce words that technically hang together, but it cannot innovate, cannot imagine, and will never write anything worth reading.
He still believes that. We still believe that. Mostly.
What we no longer believe is that we will always be able to look at an AI story and know.
Bona Books accidentally bought an AI story. Indeed, we very nearly bought two, with a further three longlisted.
And this is a problem bigger than just us.
*
How did we get here?
For a micro-press with a volunteer team, this has been a nightmare of epic proportions. Months lost, huge energies expended, new emergency procedures built from scratch. Bona Books is dedicated to promoting human authors, and will never tolerate AI-generated work. If we’d caught this even two months later, we’d have been in production and facing the very real prospect of pulping an entire print run, without money for a reprint.
Since June 2025, the team at Bona Books have been ploughing our energies into our sophomore release: Wrath Month. A scrappy, angry, punk anthology of queer speculative fiction. We hoped it would be a way to channel some of the anger that so many LGBTQIA+ people (ourselves included) have been feeling at the state of the world into something meaningful.
The community seemed to agree because stories flooded in: 606 wild, furious tales from queer writers around the world. We had magical punk rockers, trans weredingoes eating cops, zombie gays refusing to be buried. Cutting it to just twenty stories was a challenge, but we were punching the air.
One chosen story was a steampunk, anti-colonial fantasy, titled “The Machine-Breaker of Aba”. It hit our brief with a perfect intensity that felt difficult to resist. A queer woman, building a machine from scrap metal and rage to destroy her colonial oppressors? Of course we loved it. The author was an emerging Nigerian writer named Bella Chacha.
We also accepted an epic fantasy romance called “The Rot Beneath”, in which a young man merged with a mycelial network to bring down a fascist regime (and rescue his boyfriend). The worldbuilding was messy, but it offered an imaginative, epic scale that expanded the tone of the anthology. As far as we could tell, this would have been the author Stephen Jackson’s first professional publication.
(Note: We are naming both authors here because we are confident of the process we have run and able to defend the claims we have made. We also note both these names are pseudonyms.)
Much later, after months of diligent work from our tiny team, Wrath Month was inches away from going to print.
Then everything went wrong.
*
A Timely Warning
Another editor alerted us to the problem. Several SFF magazines had received numerous submissions with AI hallmarks from an author publishing under a familiar name: Bella Chacha.
Once we’d picked our hearts off the floor, we started digging: Bella had minimal online presence, first appearing in May 2025. Since then, she’d accumulated a remarkable publication record (we counted 15 published stories in 12 months), across name publishers and small or community-minded outlets, including several paying SFWA rates (like us).
That proved nothing, of course. Writers can be prolific. They can be private. A Nigerian bisexual woman (as Bella self-identified) may very well have more need of privacy than most.
So we went back to Bella’s original submission and read it again, this time with our antennae raised and with a little enforced humility.
AI-Writing Alarm Bells
Many short fiction markets, understandably, are looking for stories that are ready to publish. One way Bona Books tries to support emerging authors is by accepting stories from talented writers that still require some additional editorial support. Bella was one of these; a relatively new author (as we understood it), writing from a different literary context to most of our submissions. Turning back, many features we had previously considered distinct, took on worrying resonances.
The closer we examined that first draft, the more the story’s centre seemed absent. The protagonist was active, but lacked interiority. The worldbuilding had colourful details that somehow failed to build consistently. Certain hallmark sentences recurred over and over (“Not X, but Y…”). And the pacing was off, with important story beats given oddly little narrative weight.
Revisiting the editorial process was revealing, too. After we accept stories, we tend to send our authors an editorial letter with ‘big picture’ feedback, alongside comments and line edits in the manuscript itself. Authors engage with this discussion actively; they accept some suggestions, reject others, explain their reasons. Authors make revisions in a way that shows sensitivity and craft.
Looking over the document’s version history, we were struck by how little that had happened. Following ‘big picture’ feedback, Bella had only made one small change, without any comment. She had accepted every line-edit without discussion. And her replies to in-document comments were vacant: “thank you”, “appreciate it”, “thank you”. No questions, no usual signs of an author carefully considering a story they had created and shaped .
So we did the awful thing that needed to be done, and wrote to one of our announced, paid and contracted authors, broaching—as sensitively as we could—that maybe they were a fraud.
*
Bella Chacha’s Response
Bella’s initial response ought to have been reassuring. It was substantive and articulate, albeit emphasising that English was her second language. She affirmed the story was hers, pointing out that the story’s “setting, references, and emotional texture come from familiarity with Aba, Nigeria”, and were “not something that could be convincingly constructed without that grounding.” A statement we wanted to believe, even as it felt hallucinatory and contradictory to the manuscript. The thinness of the worldbuilding, the lack of specific references and emotional texture, were features that actively concerned us, but here Bella was pointing to them in defense.
We asked Bella to share any contemporaneous documentation she might have for writing the story (notes, drafts, messages, etc), and asked to arrange a video call, seeking to give her every opportunity for clarification.
Instead, an absurd back and forth began: Bella would promise to send evidence—only for none to appear. She would emphatically agree to a meeting, but never confirm a date or time. Over and over. For weeks. Only after putting our foot down, did Bella finally admit that actually no evidence existed.
She did agree to a meeting, though.
*
Face to Face
Which was how, several months later, we found ourselves speaking to a real life Bella Chacha. It was not a reassuring conversation.
In advance, we had prepared a structured set of topics to explore: Bella’s rationale behind certain story choices, what changed most during her revisions, her intentions as a writer, etc. Questions that would lead to a pleasant conversation with a genuine author, but that might stump someone who had generated a story with AI.
Bella gave full, if quite general, answers around the more conceptual aspects of her writing: story ideas, themes, and what stories meant to her. But as we moved on to authorial intention and craft, we kept encountering difficulty.
Bella’s answers would be simplistic, or circle back to those easier discussion points about ideas and themes. As far as we could encourage her to speak about her creative process, she gave a flattened account of ‘having an idea’, followed by ‘writing’. When pressed on matters like character design or structure, she regurgitated the anthology brief back at us.
When we specifically asked how she approached revisions, Bella began vaguely, saying she looked for bits where she “did not do very well”, before “changing” them. When pushed on what sort of changes, she claimed her biggest revision to the story had been adding the genre elements, because the first draft hadn’t been speculative at all. The Editors would like to emphasise the absolute absurdity of claiming this for a dystopian sci-fi story about a protagonist who builds a giant angry robot.
She was real, she was funny, she was actually rather charming—but we received no reassurance whatsoever that this person had written “The Machine-Breaker of Aba”.
*
The Infection Spreads
Meanwhile, we had gone back and reassessed every piece in the anthology, and found red flags around a second story. Yes, “The Rot Beneath” by Stephen Jackson.
Like “Machine-Breaker”, it had hit our brief with an irresistible energy and intensity, but contained features that now caused concern: a sense of absence at the centre; worldbuilding that was specific in detail but that didn’t cohere; oddly unvaried pacing. Again, an author with little online presence. Again, irregularities in the editorial process.
So we wrote another awful email, and while the response was different to Bella’s, it did rhyme. A long, plausible message that should have been convincing, if not for the strange inconsistencies. The only bright side was that the author had described their writing process for the revisions in detail: “I end up with 4–5 documents open, note cards scattered everywhere, and post-it notes covering everything”.
Reassured, we asked if they could share this evidence. As with Bella, we also asked to schedule a call.
The author never replied.
*
An Epidemic
Purely out of interest, we reviewed the unsuccessful stories on the Wrath Month shortlist—and found there was another.
We also turned back to revisit our submissions pile. Within the longlist we were considering for our forthcoming queer spec fic magazine, Fantabulosa!, we found two more AI-like stories.
They seemed to be emerging everywhere we looked.
We used to think AI-generated fiction would always be obvious, and we were not prepared. Over the coming years, agents, editors, and slush readers at every level are going to need to educate themselves on how AI writes.
Whilst we investigated all this, we remained silent in public. We are intensely aware of the damage a poorly substantiated accusation can do to a writer. That’s why we didn’t announce our suspicions about these stories right away. We established a detailed process, we ran it to its conclusion, and now we’re releasing our findings with as high a level of confidence as we are able to have.
*
What does this mean for Wrath Month?
This scandal has stolen months of capacity, time, labour, money, and attention away from the genuine queer writers that we exist to support. It has forced us to treat the very new voices that we aim to platform with skepticism, putting an ugly and suspicious lens over our joyful volunteer enterprise.
Every penny we lose is money that can’t go into paying our collaborators, producing books, and making radical queer art. On this front, even Wrath Month cannot contain the fury we feel at the prospect of having been conned.
So we’re going to do what Wrath Month was always intended to do: turn that fury into energy, and use it to make art. Bona Books will keep publishing queer fiction. Wrath Month is back, and now targeting publication in autumn 2026.
Above all, we are resolved that we will replace the two lost AI stories with three new human-authored ones. We are window-smashingly happy that Hailey Piper, Mimi Mondal, and Bogi Takács have agreed to step into the breach and lend their pens to Wrath Month.
*
So what?
Publishing is in crisis. One look at the improvised responses of Hachette to the Shygirl controversy, and the Granta/Commonwealth Prize schism, shows that our industry is not prepared for this. Defying all conscience, even vaunted Nobel Laureates now announce they develop stories through a personal relationship with ChatGPT.
Worse, it feels like we are barely acknowledging it—and the conversations we are having, are happening in private. There is a sense from many that publishers can sort this out quietly and autonomously, without excessive disruption. Based on our recent experience, that’s foolhardy.
The Bona Books team were at The London Book Fair earlier this year, and even as authors and artists marshalled against AI on one side of the convention hall, on the other side, others excitedly held meetings about how AI can replace human authors, artists, editors, and narrators to enhance production margins.
Professional bodies are not doing enough on this issue and should be lobbying harder. Capital protects itself, so questions around intellectual property and AI have been resolved in record speed, but where are the protections for human authors? Where is the compensation for editors and publishers who must do more work, and absorb the costs of AI-prevention measures and broken contracts? We must consider advocating for the regulatory protection of human-made art.
Most importantly, we believe that silence is the wrong response. We’re speaking openly because we believe publishers should be accountable when things go wrong. If we stay quiet out of fear, then we will continue to lose these secret battles alone, outflanked by bad actors dedicated to exploiting an ecosystem built on trust.
Stalwarts like Neil Clarke have been sounding the alarm on this issue for years. We have been waiting for a first mover to shift the dial and drive action.
What We’re Doing
Our AI Report - Alongside this post, we are publishing a detailed report, describing the process we followed to investigate AI-generated stories in our forthcoming publications, and our reflections and recommendations for other presses and the publishing industry. In time, we will seek to develop this further. We will aim to bring forward supplementary research on:
the economic impact of publishing losses due to fraudulent AI-content;
the complex implications this new editorial mode has for diversity and representation in publishing output;
updates on the mode and capability of AI-origin publishing submissions, as GenAI LLMs and their users evolve.
Our own AI Policy - We have learned an enormous amount in the last few months, and, following the reflections in our report, we will be revisiting and rewriting our own policies and procedures to ensure they are fit for purpose before our next submissions round. We will keep these under regular review as we adapt to changes in the technological threat and capability.
Anti-AI Resources - We will look to develop, and collaborate with others on, a suite of anti-AI resources for publishers. This should eventually include: a genericised version of the response procedure we have followed, detection guidance for editors and slush readers, our interview mark scheme, and suggested anti-AI wording for submissions guidelines and contracts. We are in contact with other editors who we know are already working on many of these. We will seek to collaborate on, develop, and share versions of these during the rest of 2026 into 2027.
Partner withBooks By People - This UK-based organisation is seeking to establish a ‘human-authored’ certification standard for books, involving a series of evidence-based verification steps, including a proprietary (not rights-abusing) AI checker. Bona Books are exploring how we can work with Books by People to certify our publications moving forward, and collaborate on research mentioned above.
Strengthen the Industry - Seeking to advance industry efforts and collective action to:
Bring publishing together to form a unified platform for problem-solving and advocacy on reform. Starting with the UK, with potential global reach.
Build a robust network for publishers, editors, authors, artists and creatives at all levels to examine and speak on this issue.
Identify and develop solutions through industry cooperation and unity, working to disseminate knowledge and create shareable resources.
Develop consensus on the ethics of AI-use, for publishers and authors (particularly current grey areas, such as AI Detectors and AI-powered editing tools).
We are working hard to bring forward a Bona Books slate for 2027 onwards. With these trials overcome, we’ll be working harder than ever to bring our readers an ambitious, vibratingly bold programme of radical queer literature. We’ll be aiming to shatter readers’ expectations and upend industry norms.
We hope that this insight is useful for those who read it, particularly other editors and small presses.
This isn’t a problem that we can fix alone. Community tipped us off to the problem, community helped inform the author verification process we designed from scratch, community brought us outstanding new writers to fill the gap, and community, our community, will forgive us for the prolonged slowdown as we worked out how to respond.
*
Publishers, editors, writers, readers, and artists must rise to the threat that is already inside our houses. Solidarity, humility, and just a dash of righteous rage will be essential for diverse publishing to survive.
C.L McCartney
Trip Galey
Robert Berg
Lyam Keer-Keer
June 2026
