The Real Scandal Isn’t That AI Wrote a Prize-Winning Story. It’s the Response

Recently, the Commonwealth Short Story Prize faced credible allegations that several of its regional winners were AI-generated. Granta, the literary magazine that published the stories, found itself drawn into the controversy.

The Commonwealth Short Story Prize matters. It selects five regional winners annually—one each from Africa, Asia, Canada and Europe, the Caribbean, and the Pacific—who then compete for the overall prize. Over the years, it has identified emerging authors whose careers came to justify the prize’s reputation: Jennifer Nansubuga Makumbi (overall winner, 2014), for example, who subsequently published two novels and a short story collection and was awarded a Windham-Campbell Prize.

As an author and a teacher, I have skin in this game. I was the judge for the Canada/Europe region of the prize in 2013, when it was awarded jointly to Eliza Robertson (Canada) and Sharon Millar (Trinidad and Tobago), both of whom went on to publish their books. For over a decade, I have encouraged my writing students and peers to submit to this contest, and some have made it as far as the shortlists. I did this because I believed in the prize’s mission and because avenues for new writers to get noticed are increasingly rare.

Evidence suggests that that mission is in jeopardy. For an article in The Atlantic, Vauhini Vara ran fifteen years of regional winners through the AI detector Pangram. Four were flagged as likely to have been written, in large part, by AI: three from this year and one from 2025. On its own, this doesn’t prove anything, but it makes it harder to dismiss the controversy as a false positive or as a misreading of non-western writing styles.

I’m not here to argue whether the entries relied on AI. Others (such as Wired) have done that at length. Instead, I want to draw attention to the responses from Granta and the Commonwealth Foundation, both of which seem to have mistaken uncertainty for helplessness. Whether any AI detector can deliver true certainty is not what matters. What matters is institutional fence-sitting and head-burying—some might call it cowardice—that looks a lot like complicity.

At its core, this is a debate about values. A short story implies a human artistic act with intentional imaginative labour—the exact practice whose future is now at risk if the literary world doesn’t take a stand.

Let’s start with Razmi Farook, director-general of the Commonwealth Foundation, whose response to the allegations included this sentence:

“Until a sufficient tool or process to reliably detect the use of AI emerges that can also grapple with the challenges pertaining to working with unpublished fiction, the Foundation and the Commonwealth Short Story Prize must operate on the principle of trust.”

This sounds reasonable only if we accept that the Commonwealth Foundation—and indeed, any literary institution founded on interpretative judgment, which, last I checked, is all of them—is helpless as it awaits the invention of an imagined perfect machine. But the Foundation is not helpless. It has the same tools it has always had, which still work and which literary journals and organizations use daily: analysis, comparison, critical reading, the ability to weigh evidence, to select, to not select, to take a principled stance. That is: the hard work, the human work, of assessing and appreciating art.

Granta, in an even more remarkable response, fed the accused text to Claude, asking if it was AI-generated, seemingly unaware or unworried that they were submitting (potentially) literary work to the very kind of system at the heart of the scandal. Claude is a general-purpose chatbot and not an AI detector, but it nonetheless concluded that the text was “almost certainly not produced unaided by a human.” This is not striking, but what is striking is that Granta then included Claude’s cautionary note as if it were an expert opinion, surrendering to an AI its own opportunity to pass a reasoned judgment:

“The strongest evidence against pure AI authorship is the small number of passages that don’t fit the pattern . . . Those passages carry the kind of off-shape specificity that models still struggle to produce unprompted. If the story has a human core, it is concentrated there, and the AI has been used to elaborate around it.”

It would be laughable, if not such a bad omen, that Granta thought it was acceptable to ask a large language model—operated by a company sued for the widespread use of pirated books—to draft its response to allegations of AI-generated text in a literary prize. Is there no “human core” at Granta? Surely there are people there—editors, aspiring writers—who also have some skin in this game? Why has this storied literary organization been unable to answer a literary question in its own voice?

If I sound outraged or tired (I am both), it is because I naively thought that these institutions would not go so gently into the good night. The world of letters has never operated on certainty alone; it has relied on thresholds of plausibility, exacting standards, critical judgment, and if nothing else, the underappreciated effort of people who believe that literature matters.

Integrity has long been something readers demand and publishers are expected to deliver. In 2006, for example, when it was discovered that James Frey’s 2003 memoir A Million Little Pieces was largely fabricated, his publisher, Random House, agreed to a settlement of up to $2.35 million (US) to cover refunds and legal fees. John Hughes’s 2021 novel The Dogs was withdrawn from the Miles Franklin Literary Award (Australia’s equivalent to the Giller Prize) after investigations turned up passages lifted from Svetlana Alexievich’s The Unwomanly Face of War (among others). Jumi Bello’s novel The Leaving was pulled from publication in 2022 under similar circumstances.

On a more comparable level, in 2017, the Journey Prize (the most prestigious short story award for emerging writers in Canada) faced allegations that one of its shortlistees, Richard Kelly Kemick, had plagiarized Amy Hempel’s story “The Dog of the Marriage.” Kemick hadn’t transcribed passages from Hempel verbatim, but readers zeroed in on structural and idiomatic overlaps. Both stories feature characters whose spouses are about to leave them, and who devote their waking hours to jobs involving dogs: Kemick’s protagonist uses his dog, Orville, to chase birds from airport landing strips; Hempel’s narrator trains guide dogs for the blind. In Hempel’s story, the narrator dresses a dog up as Santa to raise morale at a Catholic hospital. In Kemick’s, the narrator dresses a dog up as a pilot to ease children’s fear of flight.

Writing for the Globe and Mail at the time, Mark Medley noted that near the beginning of Hempel’s story, her narrator “[works] with these dogs every day, and their capability, their decency, shames me.” In Kemick’s story, it appears as “I work with this dog every day and every day her honesty shames me.” In Literary Review of Canada, Pasha Malla described how both stories begin with an ending and end with a beginning. Kemick’s starts with, “The last thing my wife and I did together . . .” and concludes with his characters “patient and understanding as their unclaimed futures revolve somewhere ahead of them.” Hempel’s starts with, “On the last night of the marriage, my husband and I . . .” and ends with her character waiting “with perfect hope, for the make-believe story to unfold.”

The Journey Prize cut Kemick from the shortlist. Maisonneuve, which had published Kemick’s story in its spring 2016 issue, scrubbed the story from its website and pulped what copies remained. Their then editor told the Globe and Mail that it was “a significant erosion of trust, so we won’t be publishing with Richard again, which we’ve let him know.”

That was ten years ago. Pre-pandemic. A different time. I have read Kemick’s and Hempel’s stories back-to-back and side-by-side, and I marvel, amid this unfolding AI ordeal, at the fact that the Journey Prize took such an uncompromising stance, even among chatter that the scale of Kemick’s plagiarism trended toward an unattributed homage, a misstep by an emerging writer finding his way through influence. What kind of response would we have received from Granta and the Commonwealth Foundation if, in lieu of AI generation, there had been allegations of plagiarism?

In her article for the Atlantic, Vauhini Vara quotes Farook as saying that “revoking a prize without proof is, morally and legally, no simple matter.” Sigrid Rausing, in Granta’s statement, concludes by saying, “It may be that the judges have now awarded a prize to an instance of AI plagiarism—we don’t yet know, and perhaps we never will know.”

The controversy around Kemick and the Journey Prize demonstrates that literary culture already condemns forms of appropriation that are deeper and murkier than line-by-line copying. It feels disingenuous for institutions like Granta and the Commonwealth Foundation to suddenly pretend only courtroom-grade conclusiveness counts.

• The New York Times Got Caught Using AI Hallucinations in Its Reporting
• I Broke the Year’s Biggest Literary Story. The New York Times Took the Credit
• If Chatbots Can Replace Writers, It’s Because We Made Writing Replaceable

When a text is credibly suspected of being AI-generated, it should be treated with the same righteous indignation mustered for suspicions of plagiarism: an investigation, a transparent process, a suspended celebration, and an articulation—and a defence—of integrity.

These LLMs were trained on books taken from pirated libraries (most people would call this “stolen”)—from authors, from the very community that the contributors of these texts purport to want to be a part of and that these organizations claim to support. There is no way to reconcile that; if a writer uses AI to generate fiction or a journal publishes AI-generated text as literature, they are asserting that they don’t care about the labour of the writers who’ve come before. It is an abdication of their membership in the community of writers.

The act of writing a story is itself an attempt to explore some facet of human experience, to take part in a tradition (storytelling) that is as old as we are and that, in many ways, defines us; it is the only way, however briefly, to glimpse what it is like to be someone else. Having an AI do this is to circumvent the heart of the thing.

The Journey Prize took a stance; Maisonneuve took a stance. They opted to protect the sanctity of this practice. Granta, Commonwealth: be like Maisonneuve, be like the Journey Prize.

Given these responses, Granta and the Commonwealth Foundation are only inviting more AI-generated text. This year, the prize received 7,806 entries, between 15.6 and 39 million words. How many of those are already AI-generated, and how many will be in the future? More and more. More and more. The jaded elder millennial in me wonders if these organizations are not taking a stance because the wider literary world is not yet in agreement about how much AI is too much; perhaps they are leaving the door open to future submissions done in part or with the assistance of AI generation.

Granta is (was?) one of the most respected literary organizations in the English-speaking world. Its raison d’être has been to promote literature that explores undersung aspects of human experience. They published a story of mine in 2017; it remains one of the highlights of my career. It leaves me deeply saddened that they are unable—or worse, unwilling—to articulate why (or, tabernac: that) human authorship matters.

This is bigger than one prize shortlist or one AI text published with a major magazine. That was inevitable; human beings are fallible. But it was not inevitable that the response would be to roll over and shrug. Institutions like Granta and the Commonwealth Foundation have a duty to stand up for the writers and voices—especially emerging writers and new voices—they claim to support. They are the gatekeepers; they must resist the encroachment of AI in literature.

Many of my students—excellent writers, devoted writers, mucking their way through drafts and revisions and rejections—submitted a story to the Commonwealth Prize; many of them have received form letter rejections from Granta. I tell them with all the authority of experience—as someone who participated in that ecosystem, benefited from it, sent others toward it—that this is how it goes, to shrug it off, to trust the process, to write and write and try and try. To not lose faith.

But in light of this crisis, I don’t know if I believe myself anymore. My students lost out to AI-generated text; they see those texts under Granta’s venerable masthead. And nobody with any clout seems willing to call this out as bullshit. Can I look these young writers in the eye with confidence and tell them they’ll get there if they keep on keeping on? Writing is hard. AI promises to make it easier. That allure is strong. And here before them: evidence that it works too.