Has an AI-authored story just won a literary prize? What does it mean for writing and books?

The elephant is so firmly in the room that the debate is no longer about whether Jamir Nazir’s story The Serpent in the Grove, the winner for the Trinidad and Tobago, Caribbean region of the Commonwealth Short Story Prize 2026, was in fact written by an AI agent using a Large Language Model. It is about whether a story written by such an LLM could actually convince a literary jury to give it a prize.

If that is indeed the case, the implications are extraordinary. Extrapolate from jury to general reader, and you have a situation where the very idea of a critically and/or popularly acclaimed novel or a story in the near future – or already – may well be the work, primarily, of an LLM agent.

Fundamental assumptions about the human creative impulse being the basis of art in general and literature in particular will then be seriously challenged.

The allegation

All five winning stories from each of the five regions were published online in the literary magazine Granta, as has been the practice in recent years. The first hint that Nazir’s story may have been written in part or fully by an AI agent came from writer and researches Nabeel S Qureshi, who pointed to parts of the story and suggested they were typical of “ChatGPT-generated” language.

Well, this is a first: a ChatGPT-generated story won a prestigious literary prize (The Commonwealth Prize).

"Not X, not Y, but Z" sentences everywhere, the "hums" trope, and plenty of other obvious markers of AI writing.

A major milestone for AI, at any rate...@GrantaMag https://t.co/BWGBpRasNz pic.twitter.com/U6jWejprFv — Nabeel S. Qureshi (@nabeelqu) May 18, 2026

As Qureshi cited more examples, others immediately turned to, ironically, AI to check whether the story was written by an LLM program or not. Pangram, a tool meant to detect the extent of AI-writing in any text, turned in results that declared that 100% of the text was authored by AI.

Inevitably, the possibility that an AI agent had written a prizewinning story made it to media around the world, though the focus in most cases was on whether Nazir had indeed used AI to write the story.

The LinkedIn and Facebook pages of a person with the same name and of similar appearance appeared to cement the suspicion because both are replete with observations on AI. The Facebook page also links to the Commonwealth Short Story winners’ announcement in which Nazir is featured.

More sophisticated tests followed. Computer science professor Tuhin Chakraborty used the tool Infinigram to trace the origins of some of the phrases used in the story, and found several of them in pieces of fan-fiction pieces strewn across the Internet. And for her piece in The Atlantic, writer Vauhini Vara asked Jenna Russel, a research scientist at Pangram, to test 75 stories – the five Commonwealth Short Story winners from every year since 2012 – for possible AI authorship.

Russel’s efforts showed that no stories before 2025 appeared to have been written by AI, but three from this year’s five regional winners seemed to have been. The other two, besides Jamir’s story, were claimed to have been Malta’s John Edward DeMicoli’s story The Bastion’s Shadow (winner, Canada and Europe region), supposedly 100% AI-written, and India’s Sharon Aruparayil’s story Mehendi Nights (winner, India, Asia region), supposedly 88% AI-written.

Among the four writers, only Aruparayil responded to Vara’s questions, stating that no AI was involved in the writing of her story.

We tested every Commonwealth Short Story Prize winner since 2012. We found three more AI-generated stories -- two among the 2026 winners, as well as the 2025 overall winner. pic.twitter.com/OaJbnyrJKj — Pangram Labs (@pangramlabs) May 21, 2026

The check also flagged the overall winner of the 2025 Commonwealth Short Story Contest, Saint Vincent and the Grenadines’ Chanel Sutherland’s story Descend (also the regional winner for Canada and Europe) as 89% AI-written.

Of course, questions have been asked about the infallibility of AI programs when it comes to detecting accurately whether a particular piece of text is written by AI or a human. The evidence in this regard is not yet clinching, especially as most of the accusations appear to be based on the conclusions drawn by just the one specialised tool, Pangram.

The response

The accusations have elicited responses from both the Commonwealth Foundation, which runs the short contest, and Granta, where the stories were published. In a statement, Razmi Farook, Director-General of the Commonwealth Foundation, said that the organisation is taking the allegations seriously, and will review its processes.

But she emphasised that the foundation is backing the writers, stating that declarations had been obtained from the winners that their work was their own and original, and that “...[u]ntil 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.”

In her statement, Granta publisher Sigrid Rausing said that the magazine had responded to the allegations by using Claude AI to test the premise that the stories in questions were authored by AI, and that the results were inconclusive. On social media, Granta was at the receiving end of some of the blame for publishing the story for not exercising editorial judgement. Rausing explained that the choice of the winning stories was the Commonwealth Foundation’s.

But perhaps the most significant part of the statement is the one that said, “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.”

And that is the whole point – not just for these stories, but for all writing that emerges in a world where AI tools are available to all writers.

AI is doing literature already

To many, any use of AI signals an erosion of the worth of literature. But another school of thought contends that its use is inevitable. Already, at least one Nobel Prize-winning author, Olga Tokarczuk has thrown the cat among the pigeons by mentioning that she uses AI for developing her fiction and for research.

According to a (machine-generated) translation of a portion of an interview she gave, published on Lithub, “The involvement of authors from a purely economic point of view, in this dimension of long stories, is simply difficult to imagine. Perhaps a symbiotic future and collaboration with artificial intelligence will help them. Contrary to fears, I believe that we writers, due to the specific nature of our craft, will most quickly and closely engage with tools like AI.”

The Commonwealth Short Story Prize uses a reasonably rigorous system of judging. A group of readers prepares a longlist from all the entries – which are not only in English but in 12 other languages. The five-member jury, all of them from the field of literature, then prepares shortlists and winners for each of the five regions, followed by the choice of the overall winner. The jury is not, of course, expected to test the entries for AI-authorship. They choose the winners on perceived literary merit – which, of course, can be a matter of debate.

On Lithub, the writer Innocent Chizaram Ilo wrote, “It was a beautiful story, read a bit overwritten and melodramatic in some points but this was a style of writing I am used to, written by someone Toni Morrison would describe as a person who does language.”

He added that Nazir’s story “echoes the voices of Asian and Caribbean literary canons. There are noticeable influences from writers such as Arundhati Roy and Jamaica Kincaid. It has the beautiful strangeness I always look forward to in the stories of previous Asian/Caribbean regional winners of the prize.”

He should know, since he won the African regional prize in 2020 with his story When A Woman Renounces Motherhood.

Was that what the jury thought too? According to Sharma Taylor, judge, Caribbean region, “Jamir Nazir’s language is sublime – precise yet richly evocative – conjuring vivid, lush imagery with remarkable economy. Through sharp sensory detail, he renders the Grove as a living presence, where labour, landscape, and memory are intimately entwined. Polished and confident, this is a story with a melodic voice that lingers long after the final line. Jamir Nazir’s prose pulses with a voice of restraint and quiet authority – a beautifully told and assured piece of storytelling.”

At least one commentator has raised the question of whether the kind of lush, overwrought prose of the winning story would be acceptable to the judges as high-quality had it not come from the Global South. Writing in Africa is a Country, Lina Abushouk argued that lauding this style of prose, not to mention the mystical storyline, is a reward for meeting “the enduring metropolitan expectation that writing from the South should sound opaque, excessive, and primitive.”

She said that the citation implies generic praise and does not really engage with the story in question.

Who wrote it? Will it matter?

Of course, jury perceptions are not distinct from the notions of the literary establishment. But if one or more of the winning entries of 2026 are indeed partly or wholly AI-authored, then clearly the work is perceived as being as good, and even better, than many of the human-authored entries. So the question is this: should literature now be judged on the basis of the quality of the work, irrespective of how it was produced?

If that is the case, does it signal a crisis for the human endeavour of writing? As novelist Will Self puts it, “The scale of the crisis is therefore monumental. The Granta affair is not simply an embarrassing editorial oversight. It is the first genuinely literary manifestation of a much larger civilisational fracture: institutions charged with preserving style can no longer reliably recognise it.”

This could be a crisis for readers too. As Ilo wrote, “...god forbid we unknowingly stumble on an AI-generated piece and genuinely enjoy reading it. How does one atone for this? Is it even possible to unlike prose after the identity of its non-human is brought to light?”

Then again, LLMs write what they are taught, and the texts they learn from are written by humans. When asked whether Jamir’s offending passage cited above reminded it of other writers’ works, Perplexity AI replied (the irony is not lost on us) “The passage reads most like Kamau Brathwaite, specifically from Rights of Passage (1967) and Islands (1969), part of his Arrivants trilogy. Brathwaite's method is exactly this: short, compressed units of meaning, images that carry colonial and bodily weight simultaneously, syntax that bends toward the oral and the percussive... There is also a strong Derek Walcott presence, particularly the Walcott of Another Life (1973), which began as a prose text before becoming narrative verse.”

All of which suggests that drawing a hard line between AI-authored and human-authored may be a difficult proposition.

Of course, AI-authored or -aided work may be considered to be falling short of the ethical requirement of originality we have come to implicitly expect from good art. That originality does not always manifest itself in the storyline when it comes to fiction, but in the sensation that it generates in the reader of having read something new.

Now, whether an AI agent responding to human prompts can be considered doing original work is a philosophical and ontological question – one that won’t be resolved easily.

Still, there is little doubt that the footprint of AI – which is now baked into not just Internet search results but also into writing tools on computers – will be left on all writing. Moreover, publishers are openly adding AI to their workflows. A book written by a human may still acquire the imprint of AI if the tool is used for editing. This alone may ensure that publishers intent on cutting costs and producing more books quicker – because most titles don’t sell enough to turn in a profit – will be the ones to ensure that AI becomes part of the publishing, if not the writing process.

Imagine for a moment that Nazir had not submitted this passage as part of his story, one of the several passages that a number of commentators have pointed out as bearing a clear footprint of AI.

“Wilfred’s rum-shop leaned into the road like a rotten tooth. Inside, boards blackened by smoke and sweat, the air sweet with cane and forgetting. Coins meant for rice or kerosene slid across the counter and came back white rum hot as apology. One drink opened the chest, two turned fear into courage’s cheap cousin, three steadied the hand enough to write the future in invisible ink.”

What if he had submitted this instead?

“Wilfred's rum shop was old and tilting slightly at the side of the road. Inside, the wooden walls had turned dark from years of smoke and crowded bodies, and the place smelled faintly of sugarcane. Men came in spending money they should have used for rice or kerosene. One drink loosened them up a bit. Two drinks and they felt more confident than the day had warranted. Three drinks and they grew calm enough to make plans they would mostly forget by the next morning.”

As it turns out, this version of the passage is identified, correctly, by Pangram as 100% AI-written. The question, however, is this: would the individuals who suspected the first of these passages as the work of AI also have thought that way about the second passage – which has been generated by Perplexity AI with an explicit prompt of avoiding all the obvious tics that identify a text as written by an AI agent?

We are entering an age of literary uncertainty. Fortunately, the joy of writing a novel will never be replaced by the joy of writing a prompt. And an LLM only puts words together on the basis of probability. Humans write of life and from life. Beyond the frameworks of analysis, criticism, and awards, when it comes down to everyday reading, which of them is more likely to move readers?